Driving Plant Optimization with Advanced Process Control
Once largely limited to petro and chemical plants, APC use has grown. Economic more than technical issues are key motives.
In most process plants there are one or two operators that seem to drive the process better than others. Those operators consistently create the most output at the highest quality, and can manipulate those complex interactions that make up a process. Wouldn’t it be nice if you could have them running the plant all the time?
Companies that have implemented successful APC (advanced process control) projects suggest they’ve done just that and even better. The new drivers for operation center around economic issues such as throughput levels, higher yields, lower energy costs, and generally better financial performance. How does APC operate, and how does it deliver on such lofty promises in the real world?
APC is a term that can include a range of methodologies, including model predictive control (MPC), fuzzy logic, neural networks, and others. The common objective is to find a way to manage complex interactions within a process better than traditional regulatory control. Regulatory control works best when one variable and one actuator are related. However, often moving one actuator affects several variables at the same time. This is where traditional methods lose some effectiveness.
“There are systems that do not run on just regulatory control,” says Dr. Eduardo Gallestey, ABB global business manager for optimization solutions. “The most cited examples are modern airplanes, and they really need advanced process control tools. But with most industrial practice, companies try to design processes that are stable using regulatory control. I stress the word stable, but it does not mean optimal. You can run the plant with regulatory control, but you’ll be far from optimum.”
Pete Sharpe, director of industry programs for Emerson Process Management, says that APC “can handle unruly processes that have significant interactions and are difficult, if not impossible, to operate manually. One example that I worked on was an ethylene cracking furnace where they had three different zones with one fuel. The three zones were related to temperature, but they had individual feedstocks and feed flows through each one and separate analyzers. So if you think about it from an operator’s perspective, it’s near impossible to control exactly where you want it to be. We put our model predictive controller on there and it’s able to look at how a change in a fuel valve will change all three of the compositions. It’s able to optimize and push flows right to the target for all three of those. There’s no way the operator could do that manually.”
Those that have worked with APC stress that there are processes that are complex because of the nature of interactions at work. Others are difficult to run because equipment is in poor condition or inappropriately sized for the purpose. APC won’t overcome valves that stick or unreliable sensor data.
Processes that are not stable need to keep a safe cushion between the operating setpoint and constraint to avoid exceeding critical limits. When the process is stabilized, the setpoint can be moved closer to the limit and production increases.
The word that comes up most frequently in discussions of APC is optimization. In this case it means that the system can balance all the interacting elements in a way to achieve the most desirable outcomes laid out by plant management. Often the new strategy allows things that were formerly considered tradeoffs to advance in parallel beyond points that were considered possible with regulatory control.
“There’s an implicit assumption with APC that there is an economic or business driver to the goals that the application is trying to achieve,” says Don Morrison, global product manager for APC, Honeywell Process Solutions. “Instead of looking at a problem in series, the system looks at all of the different information coming in as a single problem. It’s looking at temperatures and pressures and flows, and even transformed or calculated values based on other values, as an entire set and then solving the problem simultaneously to come up with a single solution. Depending on the type of technology, and how it’s configured, it will, in the best case, have a business or economic driver that will be something like increase feed to the plant, create higher-value products, or reduce energy consumption. Those aren’t necessarily mutually exclusive objectives. They can be implicit in a single application.”
Marc Leroux, global marketing for ABB’s collaborative production group, agrees that the program can bring multiple benefits. “One of the big benefits that people see from APC is that you’re not just impacting something like production, you’re also improving the stability of the process and reducing energy at the same time,” he says. “We often refer to it as economic process optimization, because that’s what you’re doing.”
Where does it work
When considering APC deployments, some are better candidates than others. Emerson’s Sharpe suggests four typical elements for selection: “First, I look for processes that have high interaction—if I make a move it changes a whole bunch of variables. Second, processes that are highly constrained—if you’re running against constraints, you can account for that in a multivariable control problem much more easily. Third, processes that have high energy costs—if I run a minimum reflux I can get a good benefit there, or a big production rate where a small change can have a big benefit of dollars. Fourth, high product values—where recovering another pound of product is worth a lot of money.”
System implementers agree that a process that is incapable of running with regulatory control will have to be evaluated very carefully and is probably not a good candidate. APC systems tweak controllers more frequently than human operators, and those that cannot respond precisely and consistently will normally only make disturbances worse.
Launching a project
The first step of a project is to tune up the existing instrumentation, valves, controllers, and control infrastructure to peak performance. Often when this is done, the plant will see substantial improvements without the APC, but it’s only a warm up.
Companies that are tempted to stop at that point and not make the next investment often find that those improvements don’t last. Honeywell’s Morrison has seen that happen, and says that after the tune up, “Operators see the improvements and they’re comfortable with it initially, but over time they fall back into their normal routine. As shift changes happen, one operator will come in and say, ‘Why in the world are we running here? I know the controls can’t handle this.’ They start walking things back, and then by the time we come back for the next phase of the project, the set points have moved back for no other reason than they’ve always been there. APC doesn’t have those preconceived notions. It’s able to take advantage of every opportunity that it’s given to drive optimization of the plant.”
What about the operators?
One of the challenges to APC deployments is gaining acceptance from operators, particularly since they have to relinquish control to the intelligence embedded within the system. This has to happen on two fronts: Technical issues have to be resolved, while simultaneously the company has to assure all those involved of how jobs will be handled going forward. Promoters say operators want a chance to move away from having to babysit the process. This allows time for more managerial functions related to improved output and energy consumption.
“All the employees at a lot of our end users are being pushed to consider the business aspects of decisions they make and not just the technical side,” says Don Mack, PCS7 senior marketing specialist for Siemens. “With advanced control, that’s one of the areas that we get to once we determine whether or not they’re good candidates. After that there’s discussion of how the implementation brings economic benefits in yield improvements, productivity improvements, and energy savings. Usually the operators are integral to that whole discussion.”
APC appeals to companies dealing with operator shortages and low head counts. “There’s a view that this economy is going to drive more companies toward APC so they can make due with the lower amount of labor on hand,” says Dennis Nash, president of Control Station. “If you’re a company that let people go, you’re probably thinking, ‘How do I sustain this lower head count?’ I think we’re going to find that companies find new ways of getting those jobs done, and advanced process control is going to be among the tools. The implementation of APC will replace job recovery to some extent. I have to believe that plants are planning strategic initiatives to implement systems and maintain them rather than hiring back the handful of people they let go.”
Buy external, embedded, or…
A large-scale advanced process control system requires a great deal of computing power since the system is constantly churning through extensive calculations. Consequently, early systems were typically installed separate from the installed distributed control system (DCS). Often this is still a logical choice since the APC equipment can be selected from any vendor and does not need to be from the same hardware provider. This has its own challenges in that the two systems have to be rationalized, but it avoids having to upgrade the DCS.
More current DCS platforms often have the capability to support embedded APC functions, which is particularly useful for smaller process units. This is certainly the simplest from an implementation standpoint and typically makes it the most transparent to the operators, but requires that you have a relatively recent system to find such functionality. Additionally, the same DCS processors now have to do the APC related calculations along with normal control duties, and this can compromise responsiveness depending on the complexity of the system.
…Create your own
Some companies choose to create their own APC strategies, beginning with designing their own process model. This may be forced on a company due to the extreme nature of the process involved that precludes availability of an existing solution. Others believe that making is less costly than buying.
Robert Jackson, wireless and energy product manager for National Instruments, has worked with end users who set off on this approach. “There are two approaches to advanced process control. There’s model-based control where you ask if there’s a model of how your process should perform. You use hardware and software to simulate that model against the real world and you refine your process that way. The other is where you go back to first principles, where you get the stochiometrics and the kinetics of the reaction. You look at exactly how carbon, hydrogen, and oxygen interact and react with each other.
“If you’re in chemical processing, the reality is that chemicals are pretty well defined from first principles, and your control points are typically pretty slow. The advanced portion for a chemical process in a reaction is finding how to get multiple inputs. The business side is trying to minimize the number of control points. Instead of having 10 PID loops, if you can come up with a defined way that your distillation column is going to work and set a control point at the top and bottom, you’ve got a cheaper control system. You have to ask, do you know enough about your process to start from the very bottom up, or is your process so complex and maybe so high speed that you need to have more of a mid-level mathematical approach.”
History shows that all APC projects don’t succeed. There are various reasons, but most center around two main areas. Some fail because the plant does not commitment to maintaining equipment. If field devices begin to behave erratically, the APC system’s attempts to tweak the operation from minute to minute will make disturbances worse and operators may blame that behavior on the system.
Other installations suffer from operators that simply cannot allow the system to do its job without human intervention. With its ability to press limits and work against constraints, APC can push a process into areas where operators fear to tread. When operators give into their fears, much of the benefit is lost. Systems that aren’t used can get turned off permanently and the company loses its faith in the technology.
Those that hold fast to the course can enjoy the benefits, but it does not come without a cost. The plant has to maintain top-notch performance of field equipment, combined with engagement from operators to let APC do its job. These very basic concepts make all the difference.
Peter Welander is process industries editor. Reach him at PWelander@cfemedia.com .