The chemical industry, like other manufacturing sectors, has dramatically benefited from the enormous strides being made in electronics. Many of the gains have come from allowing companies to react faster and better to changing conditions in their plants and marketplaces. Now, however, interest in smart plants, which place more emphasis on predicting rather than reacting, is growing.
Certainly, smart plant concepts such as predictive maintenance and predictive control are not new to the chemical industry. However, as Lou Cabano, president of Pathfinder LLC, an engineering firm in Cherry Hill, N.J., notes, “Smart plants have been more hoopla than tangible results so far.” But that is changing, he adds.
Tom Buckley, vice president of Aspen Technology Inc., based in Seattle, certainly agrees. “A number of leading chemical companies want to go beyond achieving better, timely response to achieving better, timely prediction.”
Significant progress in enabling technologies makes such a transition more feasible, both technically and economically. Continuing increases in microprocessor power and speed, and data storage density allow powerful capabilities to be built right into field devices. In addition, these developments have enabled analytical techniques like near infrared to move from the laboratory to the field. Meanwhile, the increasing availability of wireless devices makes getting data from remote or inaccessible locations more feasible. At the same time, fieldbus developments are enhancing communication by enabling much more information to be handled and facilitating interoperability among devices from different manufacturers.
Andrew Ogden-Swift, Southampton, England-based director of advanced development for Honeywell, speaks for many when he says there’s a lot more interest now in predictive maintenance. A May survey by Chromalox, a Pittsburgh-based vendor of heating systems, certainly underscores that. Only 5% of those surveyed now have temperature-control systems with diagnostic or preventive maintenance capabilities. However, 75% say they expect to have these capabilities in two years. (The survey also indicates that use of wireless devices will double in two years.) Doug White, vice president for Emerson Process Management, Houston, reckons that far less than half of all process plants now have any sort of predictive maintenance capabilities.
Actual implementations of predictive maintenance have led to significant gains, says White. Potential production from existing equipment typically increases 1-3% because of fewer unscheduled shutdowns, while unplanned maintenance costs decrease 10-30%. The return on investment can be among the highest of any possible plant expenditure, he adds.
Ron Durham, leader of maintenance services for NOVA Chemicals, Joffre, Alberta, says that the availability of new technology and the potential savings from improved reliability are the key drivers at his company. Predictive maintenance should figure in the plans of any company whose goal is to follow best practices, he adds.
Vendors certainly are responding to this increased interest.
In April, Emerson Process Management, Austin, Texas, extended its PlantWeb architecture by introducing what it calls a new class of smart instruments for monitoring machinery health. The initial offering, the CSI 9210, is designed specifically for AC-motor/centrifugal-pump trains. It uses vibration, motor flux, rpm, temperature inputs and embedded diagnostics to spot problems like pump cavitation and imbalance, motor electrical overloading, bearing failure and coupling misalignment. It is said to be the first such device to use the Foundation Fieldbus communications protocol.
“The CSI 9210 Machinery Health Transmitter can alert our operators in real time when equipment problems start to manifest,” says John Rezabek, controls specialist for BP Chemicals, Lima, Ohio.
Motor/pump trains were targeted first, notes Brian Humes, vice president and general manager for Emerson, because they are so common at plants and are responsible for so many problems. Emerson plans to introduce such monitoring devices for motor/centrifugal fan, motor/rotary blower, motor/gearbox/pump, motor/compressor and turbine/pump trains.
Another good candidate for predictive maintenance is the ubiquitous heat exchanger, says Kevin Fitzgerald, senior program director, new ventures, for Invensys Process Systems, Foxboro, Mass. “The key is to bring in not just vibration data, but operating information in a contextual fashion.” A model that uses such data along with temperature, pressure and flow rate can better monitor fouling.
Invensys introduced enhancements to its Avantis CM (condition monitoring) software in March, including rules-based data handling that can be used for predictive maintenance, says Fitzgerald.
Operating companies are asking for online, real-time sensing of heat-exchanger and furnace efficiencies, vibration and corrosion, and other asset-health indicators, notes Ogden-Smith. Developing new sensors for such duties might require innovative approaches to achieve easy installation and reduced costs, he says, adding that wireless technology certainly will play a role.
Expanded use of preventive maintenance doesn’t depend on new technology, though. Fitzgerald says that a lot of the opportunities will be in the basics, employing established technologies more effectively.
Getting more data isn’t enough, he cautions. Making sure that data are crosschecked and validated is an important issue. “This is not now adequately addressed or recognized in preventive maintenance. Vendors of sensors should be doing more to provide enhanced diagnostics and self-checking at the device level.” He adds, however, “Many times diagnostics require looking beyond the particular device at the broader context of multiple instruments and data reconciliation among them. Device-level diagnostics can help — and are the wave of the future — but are not the complete answer.”
Paul Kimpton, technologist, process control for Shell U.K. Oil Products, Ellesmere Port, England, foresees more and more plants using models embedded in a distributed control system (DCS) for predictive maintenance. That will require some significant changes, though, cautions Earl Ziegler, engineering technologist/specialist for DuPont in Camden, S.C. Niche players provide much of the specialized predictive-maintenance equipment and many of these vendors are not taking advantage of developments in fieldbus. Incompatibilities remain a problem. Lack of integration is a major issue, he says.
“The real challenge is to integrate information from diverse field instruments in a cost-effective manner, to pull together process and equipment data, to get a fuller picture,” says NOVA’s Durham. He does see some progress in vendors appreciating and addressing this.
Addressing abnormal situations
Unexpected incidents, such as those caused by a pump failure or a plugged heat exchanger, can not only compromise the health of equipment, but also process performance and safety. So, advances in predicting when such situations can occur promise to have implications for both equipment maintenance and process control.
According to an analysis of previous incidents, Emerson’s White reckons that improved process measurements and real-time analysis could have prevented or at least substantially reduced the damage in 25-40% of cases. Pre-empting such abnormal situations can lead to big savings for plant operators, says Ogden-Smith. Early analysis suggests benefits on the same order as those for advanced process control, he notes.
Smart devices can play a significant role in predicting abnormal situations, White believes, since they can provide diagnostics data (Figure 3). Process anomalies generally cause a predictable pattern in data that can be categorized into five distinct patterns <em dash>— representing drift, bias, noise, spike and stuck. These patterns are common for all sensor types, notes White, and can be detected by Fieldbus Foundation devices with built-in signal-processing capabilities without the need for redundant measurements, a mathematical model of the process or a mathematical model of the sensor. The high-speed sampling can spot problems that cannot be detected by traditional means, such as analysis of historian data, he adds.
The technology now has been tested at two refineries, he notes. In one case, it identified when a fluid-catalytic-cracker catalyst bed was about to lose fluidization 30 minutes before conventional measurements spotted the problem. The second trial focused on detecting flame instability in a process heater.
Both tests relied on high-speed data acquisition, statistical analysis and simple rules for classification, adds White.
Challenges in control
Model predictive control (MPC) has been around for ages, but many companies reamain leery of it. Cost historically has been a major factor. Thus, many implementations have focused on high-volume process equipment, such as at some refining units. However, MPC now makes strong economic sense for much smaller applications, particularly ones that would benefit from tighter or better product quality, says Paul Bonner, Houston-based senior director for chemicals and polymers at AspenTech. Invensys’s Fitzgerald points to recent use of MPC on multiple-effect evaporators and spray dryers.
“Before, there was a big ante to get in the game — money, time, people; now you don’t need the skilled experts,” says White. MPC today is easier and cheaper to implement, with better engineering tools for plant staff, adds Ogden-Swift.
Results can be impressive. According to White, standard deviation reduction routinely reaches 30-70% compared to regular proportional-integral-derivative (PID) control, and payback periods of a few months are common.
Improved computational speed also enhances applicability. MPC now can handle applications that it couldn’t before, such as surge control for compressors with multiple feeds and cracking-furnace temperature control, says White.
“There’s certainly a drive for more use of MPC because of advances in modeling technology, computing platforms and information technology infrastructure,” says Dow’s Gipson. The company now is in the latter stages of a rollout to broadly deploy MPC.
At refineries, the trend is to deploy MPC across related operations, notes Ogden-Swift. At Shell, says Kimpton, the latest thrust is to have MPC handle whole trains, not just units, thanks to today’s computational power.
NOVA Chemicals now is implementing MPC at plants that ordinarily might not have been able to justify it, says Tom Alloway, advanced process control leader in Moon Township, Pa. These plants are benefiting from a program that started in March 2004 to roll out MPC across the company. Already eight controllers (five at polymer units and three at petrochemical operations) have been turned on and 34 more are in various stages of implementation.
The decision to go company-wide changed the economics, says Alloway, because the corporation, not the individual site, covers a number of the major expenses.
A real turn off
But economics aren’t the only concern.
“Some companies have been scared off because of [MPC] sustainability fears,” says Yiannis Dimitratos, technical manager for process control in the Engineering Technology Center of DuPont, Wilmington, Del. “This has been an issue that has kept companies from more widely adopting MPC.”
The track record supports those fears. “The real issue for MPC is keeping operators from turning it off. Over 50% of all predictive controllers are in manual,” says Peter Martin, vice president of performance measurement and management for Invensys.
“Just putting in technology doesn’t create long-term value. There must be a plan to maintain gains,” says Gipson. “Sustainability should involve assigned responsibilities, resources and tools.” And, he cautions, “You can get into trouble if you don’t consider management of change and culture issues.”
“People totally underestimate the effort needed to keep the systems up to date,” says Shell’s Kimpton. “Half of my time is spent maintaining systems. You have to have a close liaison on a daily basis with the panel men to see how the model is doing. You must check that the model is in tune.”
“No model is accurate forever,” agrees Dimitratos. So, he’d like to see tools to identify when a model needs updating. “Every MPC vendor should be thinking of metrics for monitoring the health of the model.”
Invensys, for one, now builds technology for automatically adapting a model right into its base platform, says Martin.
There’s also room for improvement in the building of the original model. It still requires too much work, says Alloway. He’d like to see it done fairly automatically by the DCS. “Vendors can help by providing engineering tools that are easier, faster, more reliable and with better analysis capabilities to facilitate model development,” says Dimitratos. “There’s plenty of room for improvement in the user-friendliness of the engineering tools.”
Another area where vendors could help more is in the commissioning of the controllers, notes Dimitratos. Interfacing the model and the DCS does not happen automatically, but takes significant effort. Vendors should provide an engineering tool to facilitate system integration by users, he says.
A different model
According to Dimitratos, the latest advance in MPC is the emergence of state space models for controllers to replace the empirical models conventionally used. In state space models, every variable has a physical meaning. Plus, the model better matches reality, giving better estimates and leading to better — sometimes significantly better — control actions.
“This is not an incremental improvement,” Dimitratos says. The models will require less engineering effort to develop applications, thereby cutting costs, and will be easier to use and maintain, he says. However, in general, the approach will not open up new opportunities for MPC, he adds.
ABB, for one, now offers a controller using a state space model. By replacing the typical single-input/single-output step response model with a multiple-input/multiple-output state space model, its Optimize IT Predict & Control is said to more accurately predict the effect of disturbances.
AspenTech has deployed a reduced form of a state space model for MPC at a couple of test sites, says Bonner, and will introduce it in the Advanced Control Platform the firm hopes to release in a couple of months. The model trades some of the flexibility of the firm’s rigorous dynamic state space models for speed, so it is fast enough to be used for control. Potential uses include prediction of heat exchanger fouling factors and catalyst aging.
However, Dimitratos sees significant barriers ahead for state space models. “The overwhelming majority of models are not state space. Practitioners are comfortable with other types of models. So, a switch will require a culture change.” In addition, sometimes the benefits will not provide a sufficient financial driver, he adds.
The issue of integration
Farrukh Butt, manager, controls and automation, in the Somerset, N.J., office of contractor Lockwood Greene, voices a widely held concern about current, piecemeal efforts by vendors. “Everybody does their little piece. The issue is putting it all together to have a functioning system.”
The solution will not be for one vendor to do everything, says Dow’s Gipson. “My No. 1 wish is for suppliers to understand what they do well, but realize where others can contribute. It is inevitable that no one supplier can do it all. I’m starting to see an understanding of that by vendors. They are starting to partner together more and more because the competitive environment is forcing this.”
Overall, we can safely predict more and more-capable predictive maintenance and control at more plants.