Control system lifespan: How long is long enough?
At a recent control system supplier’s user group, the presenter laying out the technology roadmap said that the company is designing its systems for at least a 30-year lifespan. Is that possible, and is it a good idea? If you’re trying to determine what the practical lifespan for a control system is, you may have to wait a few more years. Many early systems, or at least parts of those systems, are still going and show no signs of quitting. With that in mind, what should your expectations be?
Let’s start by qualifying what we mean by a control system. If you consider a typical process plant, the automation consists of field devices that are wired to I/O cards, which feed to a central controller. That controller communicates with people, historians and other IT systems through a human machine interface (HMI). There are usually other smaller remote-control functions installed in the field as well, but most are connected one way or another to the central core controller.
You would be hard pressed to find much original HMI equipment installed in the 1970s, 1980s or even 1990s still in operation. Operator terminals, disk drives, power supplies, keyboards, CRTs and the like often have mechanical or electrical parts that wear out and fail. Older versions of Microsoft Windows that drive the systems are made obsolete. On the other hand, field devices, field wiring, I/O boards and core processors can and do last a long time. In fact, when we’re talking about long-lived equipment, this is typically what’s on the list.
Of course, all old systems haven’t survived. Most DCS platforms that we consider classics are really second-generation systems. Little of the actual first-generation equipment is still around because manufacturers had not developed techniques to prevent corrosion and deterioration of circuit boards and wiring. There was also rapid technical development early on. “The first generation ended because of technology,” said Ken Keiser, PCS7 marketing manager for Siemens. “There was no reason to keep the old stuff when there were so many new things available that would do a lot more. Once we got to the second generation, users were able to upgrade them in incremental ways to keep them alive.”
The business case
A major motivation for keeping old systems going is cost. Replacing equipment that runs doesn’t happen all that often. “If plant operators want to change from an old system to a new control system, they really have to have a compelling reason,” said Hesh Kagan, portfolio architect, technology innovations for Invensys Operations Management. “There are lots of ways to keep the old system going unless you can demonstrate that when you move to the new system, there’s more value than what you had in the old system. A decision to upgrade has to be values based. Certain piece parts are more expendable, more commercial and change with technologies. Other parts are more robust from a hardware perspective. The fundamental architecture, the essence of the system, is designed to be very long lived.”
So what’s out there? How much of this old equipment is still running? “We have a very large installed base of equipment that was installed in the 1970s,” said Jason Urso, vice president of technology for Honeywell Process Solutions. “The TDC2000 is still a very sizable portion. Beyond that, we have an extremely large installed base on the order of $16 to $18 billion of TDC3000 equipment in the field that was installed in the 80s and 90s. Obviously, elements of that system have been upgraded, but the core control and I/O infrastructure is still quite extensive that’s deployed in the field today. We know customers would just as soon keep that control infrastructure running and not provide any disturbance to their operations.”
Honeywell is not alone in maintaining large numbers of older systems. There are many still running from Fischer & Porter, Bailey, Moore Products and others. Few are likely to be exactly as they were when first installed, but some elements could easily be original. Keeping these systems running is challenging and requires a commitment from the OEM, whether it’s still the same company, as in Honeywell’s case, or has changed hands. For example, Moore is now part of Siemens, and Bailey is now part of ABB. One simple fact is that some of the board-level components that were common when the system was designed in the early 1980s have been discontinued. Manufacturers have to work at creating functional equivalents using newer components. The new replacement board that you’re inserting into your APACS system may not look exactly like the old one. It probably has far fewer devices on it, but it fits in the same slot and does the same thing.
Investing in the past?
This presents system providers with a dilemma. Whenever they have to look at re-engineering a board or part of a system because of component-level obsolescence, the question emerges if this is necessary or if it is time to push users toward something more modern. “You can only do that so long before you ask how much more you’re going to invest in this product,” said Keiser. Siemens still supplies equipment for old Moore Products systems, but in some cases tries to steer customers toward its current PCS7 offering.
“For example, we upgrade the controller cards in APACS,” Keiser said. “But at some point you ask, do we continue to invest in R&D on the old equipment, or do we invest in a migration path so that you can get to PCS7 in an easy way? The development cost may be the same for that as developing a new controller with new chips for the old system. You might as well make an exit strategy for everybody. We’ve been investing in APACS, but we’ve also been investing in the exit strategy.”
Creating an exit strategy involves finding ways to improve system performance and reliability while minimizing the amount of new hardware and wiring changes. When suggesting a migration strategy, suppliers will go to great lengths to find ways to retain as much of the existing infrastructure as possible. Users don’t like to be forced into upgrades simply because the original manufacturer has decided to pull the plug.
Is long life desirable?
Haven’t there been many technological advancements since these systems were installed? Doesn’t keeping an old system retard your ability to embrace things like smart instrumentation and asset management programs? Some of that is true, but system builders have found ways to mitigate the problem.
Most advancements either interface with or are included in the HMI layer, so upgrading that facilitates many improvements, including:
Integration with enterprise level IT networks
Cyber security strategies
Data collection and historians
Sophisticated operator display graphics.
At the same time, anything that requires improvements at the I/O level is effectively locked out. That means you may have to find creative ways to get around a problem or create the means for an upgrade. For example, let’s say you want to implement an asset management program that uses HART and other features of smart sensors. If you have dumb analog I/O, what can you do?
“We do a lot of asset management projects on the older equipment,” said Urso. “The approach is that we use a HART hardware multiplexer that can be installed pretty simply in the control or instrument rooms, and it strips off the signals that run over the 4-20 mA wiring. With a hardware multiplexer and our asset management software, those customers are good to get up and running with an asset management program.”
Having said that, Urso points out that asset management isn’t what most customers are looking to add. Most of the time, there are other items on the upgrade wish list with higher priority. “Customers focus on two areas,” he said. “With a legacy system, the first question we get is: ‘I’d like to keep all of my control and I/O infrastructure in place, but I need to try and squeeze a little extra throughput out of my operation, or drive down energy costs. Can I put APC (advanced process control) in place?’ It’s important to have a bridge between legacy systems and advanced applications that generate substantial benefits.
“The second place they look is the HMI side. They know that 30% of abnormal situations are a result of human error. It’s in a context that the person didn’t have the right information to make a good decision, or they made an inadvertent mistake. There’s a strong belief that we can drive additional throughput and substantial reductions in unplanned capacity losses by improving the HMI and implementing capabilities such as procedural operations to automate some of those manual procedures. Once those two bases are covered, they tend to look at asset management.”
Buying old, new
While some old platforms have been completely phased out by their suppliers, some are still fully supported. Some suppliers can still deliver a brand-new version of the older platform, but they may not want to. Keiser said he still has to deal with those questions. “If there’s a customer that has an APACS system and they want to expand the plant, we would support that, but at the same time we would explain to them all of their other options,” he said. “If they’re adding a boiler and saying ‘this is the first one and we’re going to add 10 more over the next 20 years,’ we would strongly suggest that they not keep the APACS system. They should go with PCS7 for that. If you’re looking at a long-term expansion strategy, we think this is a better approach for the long term. We’re not going to sell an APACS system to a 300 million gallon per day wastewater treatment plant that they’re building from scratch tomorrow.”
Will the new systems installed today last as long as the classics? The expectation is yes, and probably even longer. Kagan said system design strategy has evolved over the years: “At this point in our lives, anyone building a control system should be building it in such a fashion that it continues to be upgradable, and migrates with the technology as technology changes. It should last forever, recognizing the fact that the core, the essence of it, is becoming much more open and standards based. Piece parts wear out and get replaced but they get replaced with parts that in some way are interoperable with the older pieces. The way we, as system builders, differentiate ourselves is how well we execute that and stay true to those architectural principals of allowing integration and interoperability of the parts as they change over time. Moving forward should be much more fun than it was in the past.”