Development opportunities in process automation: process automation research can lead the way to sharply reduced manufacturing costs and better decision making.

Editor's Note: The Forest, Wood, and Paper Industry Technology Summit held this past May included wide-ranging discussions on research opportunities that may produce major benefits for the industry. As one of the nine working groups, the process automation group addressed technology gaps related to sensors and process control systems. The group's invited participants, shown in the box below brought a diverse skill set of experience and knowledge to the sessions. TAPPI will summarize and publish the groups' discussions in the near future. Summaries off all sessions are available at www.tappi.org/ctosummit.asp.

Over the past 30 years, the process automation has dramatically changed the operation of pulp and paper mills. Tighter control of critical processes has reduced costs while producing important productivity and quality gains.

Yet the revolution is not complete. While process automation has produced a wide range of benefits for the forest products industry, many improvement opportunities remain. Available control technologies are not being used to their fullest potential, and other industries--notably petrochemicals--seem to have more effective process control systems.

The Technology Summit process automation working group developed a vision for process automation research and program goals that align with more general Agenda 2020 program goals, as shown below. To address these ambitious program goals, the participants identified technology gaps in process automation and possible research programs in three focus areas:

1. decision-making tools

2. process visibility/sensors

3. control system techniques.

PATH FORWARD

Process automation discussions at the Technology Summit identified many research areas that may benefit the forest products industry. Development of new sensors and use of new sensor technologies can enhance awareness of process status. Improving process understanding through first-principles dynamic models, mining information from available plant data, and enhancing the ways employees assimilate plant information will enable faster and better decision-making. Developing new or improved techniques to improve control system effectiveness can provide large rewards.

The forest products industry will continue to improve its capabilities in process automation. The rate of improvement can be increased by using resources such as DOE Agenda 2020 support to fired worthwhile research programs. The path forward requires us to locus on converting the promises of new technologies into commercial reality.

ACKNOWLEDGEMENTS:

The process automation work group members at the Technology Summit did an excellent job of identifying notable opportunities. Their contributions in defining the gap-filling technologies and working cooperatively are gratefully acknowledged.

Participants in Process Automation Work Group

PETER ARIESSOHN Weyerhaeuser, Industry--R & D

LARRY BANTA West Virginia University, Mechanical Engineering

RON BROWN * Westvaco, Industry--R & D KAREL CERNY * Georgia Pacific, Industry--Engineering

FRANK DOYLE University of Delaware, Chemical Engineering

ELMER FLEISCHMAN Idaho Nail Envir & Eng Lab, Government--R & D

ROSS MACHATTIE Honeywell Supplier--Product Development

MIKE PEMBERTON ITT Goulds Pumps, Supplier--Marketing

MIKE WALLACE Mead, Industry-R & D

STEVE WILLIAMS Aspen Technology, Supplier--Corporate Development

LYNNE MAYS TAPPI, Session Facilitator

* Co-chair

PROCESS AUTOMATION VISION AND PROGRAM GOALS

VISION

Enhance competitiveness through innovation in sensors and process automation to:

* Reduce manufacturing costs per unit of product

* Enable higher effectiveness in decision making

PROGRAM GOALS (aligned with Agenda 2020 Goals)

Use process automation to significantly cut unit manufacturing costs by:

* Reducing product not meeting first quality by 50%

* Reducing costs for materials, energy, labor, and maintenance by 10%

* Increasing production rate without significant capital investment

Enhance effectiveness of decision making by:

* Using resources optimally

* Integrating enterprise systems

* Providing training systems with realistic process simulations

* Improving operator ability to handle abnormal situations

FOCUS AREA 1: DECISION-MAKING TOOLS

Developments in process understanding and information management can enhance the ability of employees to make informed decisions quickly,

Gaps: There are several important gaps for decision-making tools:

* Inadequate process understanding is a hindrance to effective control in many process.

* Dynamic process models that are adaptable and can be extended over the full range of operating conditions often do not exist.

* Information presentation to operators is restricted by the limitations of CRT displays and control panels.

* Relationships among process variables and information about long-term dynamic responses are not presented to operators.

* Employees get diagnostic information from control systems largely upon demand.

Gap-filling research areas: The group suggested three areas for research on decision making tools.

First-principle dynamic models: The interactions among operating parameters and cause/effect relationships within many processes in the forest products industry are not well understood. Engineers must recognize certain fundamental scientific principles to develop a robust dynamic process model covering the full operating range. Once the model is validated, operators and automatic control systems can use this dynamic model in decision making. Many available first-principle models provide only steady-state predictions (based on unchanging operations), although dynamic capabilities are needed (based on "real world" situations where operations change). The dynamic aspects require considerably more attention during model development than steady-state issues.

The Summit group recommended that dynamic models be developed in two categories--to understand processes that are new to the industry or not well understood and to estimate product properties and process parameters.

Black-liquor gasification is an example of a process new to the industry for which first-principle dynamics models are not available. Dynamic models to estimate product properties and process parameters for pulping and sheet forming, as another example, provide an effective way to obtain information for decision making. Few first-principle models in forest products processes have enough sophistication to allow their use in this way.

Data mining/preprocessing of sensor data: A wealth of unused information exist in process areas in the form of raw data from sensors, control valves, drives, pumps, and other elements of a process. Researchers are developing data milling techniques that monitor available data from various sources and identify useful information about operating conditions and pending upset situations. Another approach uses sensor data to determine information beyond the sensor's primary purpose, a technique called preprocessing of sensor data. Information from these techniques can be used for asset management and real-time process control.

Data presentation techniques: A principle challenge in process control is to present information to plant personnel rapidly and effectively so that sound decisions are made quickly. Improved methods for presenting plant data and process conditions to humans are desired, especially for abnormal situations. Operators, supervisors, engineers, maintenance personnel, and business managers all could benefit from more effective presentation of process data.

Research needed: Considerable research has been done in the topics above, but mills need commercial-scale development and application to the specific requirements of the forest products industry. Universities and national labs are the likely leaders in developing new decision-making technologies.

FOCUS AREA 2: PROCESS VISIBILITY/SENSORS

Various groups have developed lengthy lists of desired sensor developments, such as the 1996, 1992, and 1988 TAPPI Research Needs Conferences. Discussions at this year's Technology Summit also generated a list of sensor needs, although it was not intended to be comprehensive. The group recognized ongoing Agenda 2020 projects for measuring recovery boiler variables, fiber properties, and full-width web properties as potentially valuable, but did not list these sensor targets as "new" needs.

Gaps: Current sensors are inadequate in a variety of respects:

* Sensors are not specific or sensitive enough.

* Sensors do not exist for some purposes.

* Some sensors are not rugged enough for on-line use.

* Sampling methods are not sufficiently accurate and introduce errors.

* Existing sensors are inflexible.

Gap-filling research areas: The Summit discussion group suggested the sensor development opportunities listed here in italics.

Miniature wireless sensors that can flow in a process stream or travel inconspicuously inside a product offer opportunities for new insight. For example, a miniature sensor that flows once through or circulates in a mill process could measure pressure, temperature, flow rate, or chemical composition locally along the sensor trajectory. A device about the size of a pulp fiber embedded in a paper web could be used to evaluate pressing, drying, calendering, or coating, and could provide information about performance during conversion or use.

Low-cost wireless sensors offer opportunities to add sensors where wiring is difficult or where the costs could not otherwise be justified. For example, sensor installations on rotating equipment such as dryer cans would be challenging without wireless technologies. Another promising application would be using sensors in remote process areas.

Biological or biochemical sensors can be specific and sensitive to individual chemical components in a process stream. However, biosensors are largely being developed outside the forest products industry.

Z-direction sensors could determine Properties through the thickness direction of a wood product or a moving web of paper. Profiles of moisture, density, filler, temperature, and fiber orientation could be useful in manufacturing products with enhanced end-use capabilities. Exploratory research in this area with magnetic resonance techniques has shown promise.

The industry needs low maintenance non-sampling stack measurement systems certified by regulatory agencies. Current approved methods are expensive to install and maintain. Research should pursue laser, ultrasound, and/or FTIR technologies to measure component gases, including total reduced sulfur and volatile organic compounds (VOCs), and to determine stack opacity and flow rates.

Rugged, in-line liquor sensors could measure concentrations of species in pulping liquors. White liquor and green liquor principally contain sodium salts. Black liquor contain inorganic salts and organic components. Mill can use real-time knowledge of these liquor components to reduce pulping variability and to achieve higher chemical recovery efficiencies. Spectroscopic techniques for these applications are promising. Auburn University has reported a mill demonstration of a near-infrared system for on-line liquor analyses.

Kappa number--the lignin content of pulp--is difficult to measure in real time. A rugged, in-line Kappa sensor to provide continuous Kappa measurements would be useful. Available measurement systems are not rugged enough and require too much maintenance.

Inexpensive diagnostic sensors for short-term use that do not require the durability of conventional sensors would be useful. The group knows of no current work on low-cost, disposable sensors, although examples of more expensive diagnostic sensors have been reported.

Sensor arrays throughout a building or process area to measure sound, vibration, and temperature could benefit pulp and paper mills. Technology is available, but demonstration of its usefulness in the forest products industry is not known.

Mobile robotic sensors for vibration, noise, temperature, and odors could act as "operator assistants" in situations where stationary sensors are not practical.

Research needed: Basic research is required for some of the sensor developments suggested above, while technology for other applications has been developed and needs to be transferred to the pulp and paper industry. National laboratories, universities, and suppliers will lead the development of sensor technologies. National laboratories have significant capabilities in sensor technologies, as demonstrated by Agenda 2020 projects underway at the Oak Ridge, Pacific Northwest, Lawrence Berkeley, and Lawrence Livermore national laboratories.

FOCUS AREA 3: CONTROL SYSTEM TECHNIQUES

The forest products industry uses state-of-the-art control system as well as many legacy systems. Most systems installed since 1980 are based on digital technologies and many are sound platforms to implement new techniques that can enhance control system performance.

Gaps: There are several important gaps in control system techniques:

* Existing control systems often do not use advanced methods such as model-based control.

* Good control performance requires continual loop tuning, but the self-tuning approaches evaluated at many mills have been inadequate.

* Automated diagnostics currently in use are relatively basic and information to operators about the control system performance is weak.

* Operator confidence in automatic control systems is frequently low due to experiences with poor loop tuning of faulty sensors. Consequently, operators often run critical control loops in manual mode rather than automatic, and product quality and process performance are more variable than they should be.

* Model development methodologies are not available for combining the best features of analytical models with those of empirical/statistical models.

* Changes in product mix and feedstock are commonplace in forest products operations, yet control of transitions in major process areas typically is accomplished manually rather than automatically. Production changes ripple through the manufacturing chain causing upsets and products that do not meet specifications. The time to reach a new steady-state operational level is greater than necessary.

Gap-filling research areas: The industry would benefit from improved techniques for developing and using advanced control methods, as follows:

Self-diagnosing, self-tuning control techniques: These control systems should be fail-safe and capable of operating with minimal operator intervention. Control actions should adapt to grade changes, minor process modifications, and sensor and actuator malfunctions. Development is needed to make self-tuning algorithms more robust and to integrate diagnostic capabilities into control systems.

Transition control: Mills need a dynamic, predictive control system that coordinates multiple processes. This system could reduce costs, produce higher levels of first-quality products, and decrease time-to-steady-state-operation after significant transitions. Aspects of the desired system are available or are being developed. Much of the development in this area will be applicable to all continuous process industries, not just forest products.

Hybrid analytical/empirical modeling: Analytical, first-principles models are often too cumbersome for real-time feedback control. However, they can be extrapolated to conditions beyond the basis of the model. In parallel with first-principle modeling efforts, powerful new techniques have emerged in recent years for generating empirical models based on process data, such as neural network methods. Empirical models can describe complex processes with interactions among variables, but only over the range for which data have been used to develop the model. Therefore, extrapolation to new conditions is not advisable. A framework and methodology for combining the two modeling approaches is needed that exploits the good features of both and provides a flexible and more powerful class of dynamic model. Many industries would benefit from this work, so its development does not depend solely on forest products support.

Process control/business systems integration: The widespread implementation of enterprise resource planning (ERP) systems in forest products companies offers opportunities for improved business efficiency. Full integration of business systems and plant process control systems has not been widely implemented, although other industries, such as petrochemicals, are more integrated than forest products. ERP links with plant process control are essentially adequate and do not need technology development. Demonstrations can help showcase the capabilities of these systems in our industry.

Research needed: The control system enhancements outlined above require applied research, commercial-scale development, and technology transfer. The greatest need is for transfer of developments from other industries into forest products using showcase demonstrations to prove their commercial value. Organizations likely to lead programs for these enhancements are universities and national labs. Supplies will be involved in commercial-scale development and technology transfer.

G. Ronald Brown is research director at Westvaco Corp in Laurel, Maryland, USA; email [e-mail]