Procedure automation, state-based control, procedural operations… Whatever you call it, automating state changes within continuous operations is just good practice to improve safety, productivity and repeatability. Some in industry have been automating for decades, incorporating batch programming into continuous processes to deal with abnormal situations. Others are only recently seeing the light, having steadfastly adhered to manual procedures to move work from one state to another.
Though it has been around in one form or another for many years, there is a more recent move to formalize the adoption of state-based control in process industries. At the forefront of this movement is the ISA-106 standards committee on Procedure Automation for Continuous Process Operations, formed in 2010 by the International Society of Automation (ISA, www.isa.org) to standardize methods to automate procedures for moving from one state to another in continuous processes.
State-based control is essentially abnormal situation management—being able to respond to that abnormal situation, whether it’s a startup that happens every seven years or some other break in the continuous process, notes David Funderburg, technology manager for the chemicals, oil and gas business for ABB (ABB Inc., www.abb.com). “Automation makes it consistent, reducing operator intervention,” he says.
One example of how continuous process uses state-based control could be in a sequence of events to clean tanks, notes Tony Barrancotta, engineering manager for the PlantPAx development team at Rockwell Automation (www.rockwellautomation.com). That sequence could include steps for a rinse, adding caustic, flushing with caustic, adding acid, flushing with acid, and a final rinse.
But it’s not so easy to automate the entire cleaning process because that process could change at any given time. “They might use all of those states or just a rinse state, depending on what kind of cleaning they want,” Barrancotta says. “They might want to rinse five minutes instead of 10 minutes. They might be looking for a certain pH. They might have to do multiple rinses; or rinse, check a lab sample, and rinse some more.”
To keep it flexible, manufacturers have often avoided automating those procedures, Barrancotta says. Instead, they’d have an operator manually opening valves here, closing valves there, turning on pumps, etc., following standard operating procedures(SOPs) to run through the process.
State-based control is an effort to define the steps to make it easier to automate. It borrows concepts from batch control to help continuous processes, “enforcing rules about moving from where I am to where I want to go in defined steps,” Barrancotta says. “They’re things we take for granted in batch. We can hold. We can start. We don’t think about those things in continuous process.”
Just good practice
Lou Bertha doesn’t see what all the hubbub is about. As a system integrator for a variety of process industries, he considers it his job to automate wherever he can, including areas that some of his customers have been hesitant to automate in the past.
Bertha’s company, RDI Controls (www.rdicontrols.com), has been involved in several gas turbine retrofit projects lately, which Bertha describes as a self-contained system. “The operator gives it a start, and the auxiliaries sequence through,” he says. The operator interaction involves starting it up and telling it to go to a certain point, then walking away. “That’s what we are typically providing.”
In many cases, automating state changes is essential as a safety measure. Controlling the flow of natural gas into the turbines or jets, for example, there are times when an operator will want to force outputs on the system because of troubleshooting or testing, Bertha notes. RDI’s automated controls allow that, but keep the operator from being able to open both upstream and downstream valves at the same time.
There have been times, particularly during startup, that a new operator is checking stop valves, and tries to check another valve while the first one is still open. The operator complains that the system mistakenly kicks back to auto mode, but in fact that’s exactly what it’s supposed to do. “Certain combinations are not allowed,” Bertha points out. “No one wants to see full gas going through a jet when it’s not operating. Things tend to blow up when you do that.”
As someone whose business it is to add new controls or retrofits for manufacturing customers, Bertha takes automating such procedures as a given. “To me, this is standard operating procedure on our systems,” he says. “We take all that into account. The worst thing that can happen is liabilities.”
Moving beyond manual operation
But not all manufacturers take that level of automation for granted. Though advancing years have brought advancing automation to continuous process industries—including remote actuators on valves, PID controllers, and into distributed control systems (DCSs) and centralized control rooms—the fact remains that most continuous process plant operators still perform most procedures manually, according to Dave Emerson, director of Yokogawa’s U.S. Development Center (Yokogawa Corporation of America, www.yokogawa.com/us).
Presenting at a Yokogawa Users Group conference, Emerson noted that key candidates for state-based control include complex procedures such as starting a distillation column; or repetitive procedures where operators add little value, such as starting pump sets by controlling individual values and pump controls.
Again, safety is key in both types of scenarios. At the Kern Oil Refinery in Bakersfield, Calif., for example, workers overpressurized a pump casing during a crude unit startup. The casing ruptured, releasing and igniting hot oil that immediately exploded, killing one employee and injuring others. At the Giant Industries Ciniza Refinery in Gallup, N.M., six employees were injured when during HF alkylation unit maintenance a shut-off valve was not closed as required, causing the release of flammable liquids and vapors that subsequently exploded.
Automation of those procedural transitions would have avoided such catastrophic situations. Evonik Industries, for example, improved process safety margins significantly when it automated procedural operations in it acrylic acid production. And safety wasn’t the only benefit. Startup time to steady state was reduced by 30 percent, and reactors came on stream 70 percent faster. Overall, there was less variability in startup time than there was with manual operation.
The same was true for Chevron when it introduced “transition automation” to a base oil hydroprocessing facility, finding consistency as a key benefit. Not only that, but consistency to the highest level, which translated into 30 percent faster transition times on average and increased product yields during transitions.
Process automation technologies in U.S. refineries are limited primarily to safety systems and simple sequential automation, often optimizing a unit only once it is in a steady state, according to Robert Tsai, a process control engineer for Chevron, in a presentation at an AIChE meeting. A refinery transitioning from one process state to another—adjusting unit operation for product grade or feedstock quality changes, for example—depends on operators to perform those transitions, he noted.
In any such case of manual operation, there is variance in human performance, no matter how well SOPs are laid out. Operators must make sure they follow each step correctly.
“With the SOP, operators have to make sure they go back and do all those steps correctly,” Barrancotta says, referring to the example in which a lab sample must be completed before continuing with a cleaning process. With state-based control, the process will automatically go back to the state model and continue to operate.
“It gives me a way to make it pretty simple for the guy coming in,” he adds. “I can give them one button, putting all that knowledge into the control system.”
This is important in times of high stress or outages, ABB’s Funderburg notes, when manual procedures aren’t always followed to a T. “But if it’s part of the automation philosophy, then it’s enforced by nature.”
But it’s also important in everyday operations. “We collect that history from operators,” Funderburg says, “so new operators can be as productive as their predecessors.”
Chevron’s hydroprocessing unit produces various grades of base oils over the course of a month, requiring frequent feed and product changes on a feed wheel. Although multivariable control optimizes each production run, there was an opportunity to optimize the eight-hour transitions between product runs, where each operator would perform those SOPs slightly differently, resulting in varying transition times. The automation that Chevron added to close the advanced process control gap operates comparably to the best operator performance, providing the improved transition times.
The solution—a simple, graphically based instruction logic to execute complex programs over an OPC connection—is not fully automated. It still requires an operator to execute some tasks in the field, such as routing tanks or stopping and starting non-automated pumps. But it significantly decreases the number of manual moves required by an operator, letting him focus instead on other parts of the plant.
A typical manual transition requires about 300 board moves in the eight hours it takes to complete a feed switch, according to Chevron’s studies. Many of those moves are lumped in a four hour period in the middle of the transition, placing even more burden on the operator. By using the automation program, that operator load drops by about 60 percent.
And, because the automated system makes about 3,000 moves compared with the operator’s 300, it is a more refined solution, providing a smoother transition in less time.
Making it easier
Although the ISA-106 committee is working on formalizing methods to automate the move from one state to another, the ability to automate these processes has existed for decades.
Engineers could create custom programming to perform batch-like functions on a more automated basis.
The problem Barrancotta saw with this was that manufacturers often needed to hire a consultant to help automate. “They would go in and create a lot of custom programming. And then any time there’s a change, they have to call a consultant again,” he says. “It’s time-consuming and expensive.”
RDI Controls and other system integrators take care of that programming for their customers. Others might try to handle it in-house.
Chevron chose to avoid making operators deal with programming code for its automation efforts, making the application instead almost completely graphical so it could interface directly with the operator. “This results in decreased time between revisions and increased operator acceptance,” Tsai noted in his paper on the subject.
Rockwell Automation’s goal is similar with its PlantPAx program, providing process objects to enable the automation to be done from the HMI. It’s easy enough to write out process requirements and specifications, Barrancotta notes, but to turn that into code in a controller is challenging.
It’s always been possible to write the code for state-based control, Barrancotta adds. “To me the real game changer is to do that in the HMI; do it visually, in a way that anyone can understand,” he says. “It’s always been possible to write a program. It has not been possible to be flexible from the HMI.”
A graphic approach works for RDI Controls as well. As Bertha points out, engineers could open up the I/O and force the outputs anytime. “But what we had to incorporate into our strategy was a reason not to do that; a reason not to go into the actual logic,” he says. “We want to do it through a graphic do it through another means so that it would keep things in a safe scenario.”
In many ways, the key to developing state-based control is an overarching strategy. “It’s not something that’s sold; it’s part of good engineering,” Bertha says. “It’s a good approach to your process.”
Funderburg agrees. “It is a design operational philosophy that really crosses how a gas facility operates a facility,” he says, explaining that ABB is involved from a control perspective on continuous improvement, to collect information from operators or engineers and put it into operation. “A lot of the work that we’ve done is to facilitate the process of spending upfront design time to put controls in to make it smarter and smarter as time goes on.”
Some customers want to take the philosophy even further. Rockwell’s next move is to provide a sequence of objects within the programming. “We’re looking at how we could tie those individual sequences together, along the lines of batch executives,” Barrancotta says (batch executives manage sets of equipment and recipes in batch processes). “We’re currently developing a flexible state machine that knows what to do at a procedural level. It ties it all into one state model, and it’s highly configurable as well.”
“We could write a sequence that does the rinse, acid, caustic, and tie them together to to do a complete clean of the vessel,” Barrancotta adds. “It’s a phased approach. So first we automate the procedures. And in the future, we could tie those sequences together. It’s another level of abstraction, with one button that says ‘clean machine.’ It would be able to monitor the individual sequences, and it would know where to take it next based on the results.”
This kind of progression is based on feedback from customers who are asking for a higher-level state-based machine with more batch abilities. “It’s getting closer and closer to the ISA-88 models,” Barrancotta explains. “They’re really seeing the benefits of ISA-88, and want to extend those concepts.”
ISA-88, though widely adopted by the batch industry, has not been a great fit for continuous process, requiring tweaks. Thus the creation of ISA-106 as an attempt to get continuous process industries to standardize procedure automation, Emerson explained.