Profit from the Loop

How control systems engineers are using loop monitoring and performance supervision software to generate measurable economic benefits for the plant

There was a time when “control loop optimization” meant selecting the proportional, integral, and derivative tuning parameters for all of the PID controllers in a plant so as to optimize the closed-loop performance of each loop individually.

Loop tuning is still a critical element of loop optimization, but thanks to recent advances in control theory and commercial software products, the control operations of an entire plant can now be optimized to maximize overall profitability, not just individual loop performance.

As George Buckbee, Vice President of Marketing and Product Development for ExperTune (supplier of PlantTriage software) puts it, “Control systems personnel are using control loop monitoring and performance supervision software to generate measurable economic benefits.

“Today’s loop optimization software focuses on providing meaningful, actionable metrics that deliver results to the bottom line and graphical tools to focus the users’ attention on the most economically important issues, thereby helping them to focus their time and effort on the areas where they can make the biggest improvement.”

Dr Alireza Haji-Valizadeh, Manager of Technology Development for ControlSoft, adds that today’s loop performance monitoring and diagnostic tools can:

• Provide both automatic and manual loop tuning and analysis tools
• Keep track of all the loops in the plant no matter what platform or vendor
• Execute non-intrusive, continuous on-line performance analysis
• Generate customizable loop performance indices such as error distribution, integral of absolute error, process variability, and economic impact
• Disseminate loop performance reports to all levels of the plant’s business hierarchy and across the Web

Maintenance & more

Haji-Valizadeh notes that his company’s InTune software (version 5) offers all of these features as well tools that are capable of addressing maintenance issues such as:

• How a loop’s performance foreshadows future maintenance requirements
• How loop maintenance efforts should be allocated to maximize return on investment
• Which of the problem loops have the most room for improvement
• How to keep loops continuously optimized on a plant-wide scale

Meanwhile, Steve Johansen, Global Services Marketing Manager for Advanced Solutions at Honeywell Process Solutions points to Honeywell’s Loop Scout as a comprehensive loop management workflow solution that contains loop analysis tools, valve diagnostics, a robust control theory based tuning package, and alternative techniques to PID control.

Loop Scout collects data on process values, setpoints, controller outputs, and control loop configuration data throughout the plant; then generates performance reports for each loop. Expert guidance tools walk the novice user through the data and offers additional tests to determine if there’s a maintenance issue (such as stiction in a valve), a tuning issue, or a problem with the control strategy itself.

“Loop Scout has evolved to the point where the user doesn’t need to understand the underlying technology. He only needs to drill down to the specific information that impacts decision-making. The user can choose which characteristic by which he wants to rank and sort his loops – the severity of each problem detected, the general performance of each loop (good, fair, poor, etc), or the greatest opportunity for improvement in terms of business metrics,” says Johansen.

John Caldwell, DeltaV Product Marketing Manager at Emerson Process Solutions adds that DeltaV InSight is an online process modeling tool that automatically learns the process based on normal day-to-day operations then monitors performance, and identifies and diagnoses problem loops; it also recommends tuning and maintenance improvements.

Better, but not perfect

Whether due to advances in loop optimization technology or other causes, there has been some improvement in control loop performance across the process industries. The average respondent to a 1996 Control Engineering US survey reported that approximately 55 percent of their control loops were running in closed-loop rather than manual or open-loop mode. Of those running in closed loop, 45 percent were believed to be optimized.

More recently, a survey of PID controllers conducted by Honeywell Process Solutions (see chart) showed that 64 percent were operating in closed loop with 50 percent of those tuned to provide “excellent” or “acceptable” performance – which still leaves a lot of room for improvement. Yet, loop optimization tools aren’t always the best way to realize additional controller performance.

“I once used a commercial loop tuning package and didn’t find it very useful for the particular loop I was trying to tune,” says Matt Lyles, Managing Partner at SBL Systems, a control system engineering company. He suspects that the significant deadtime in his loop may have been the problem, so he resorted to a tried-and-true manual tuning technique.

“I can see where there could be some critical processes out there where there is limited time allowed to tune a loop and where either a good loop tuner or derivation would be necessitated,” Lyles added. “Typically the loops I have worked with were not critical and allowed me plenty of time and even wasted product to get the loop right.”

Process control consultant Barry Payne, of Barry Payne & Associates, also has mixed feeling about loop management software. “We have found that tuning software is useful if the control loop it is applied to is functioning properly, and the person using the software already understands what a ‘normal response’ is for the process.

“If the loop is subject to significant unmeasured disturbances while the tuning software is being used, or if one or more of the end devices is malfunctioning, then the tuning software (at least that supplied by a few of the larger DCS manufacturers) may give unpredictable and frequently erroneous results.”

He cites an example where a DCS-based automatic loop tuner failed to recognize that the very low model gain it had determined for a particular process was due to an instrument malfunction rather than the actual process dynamics.

“Anyone with experience tuning this type of simple loop would have quickly recognized that there was a fault. Auto-tuning software can be useful, but only when used in combination with experience in tuning a particular type of application. It is not an ‘easy button’ for those with little or no skill at loop tuning and troubleshooting.”

========================================================================

Teaching an Old Mill New Tricks

Being able to identify poorly performing control loops among the vast number in this pulp and
paper mill was a key step in improving product quality and reducing plant downtime.

Matrikon Control Performance Monitor’s continuous monitoring capability provides control engineers and maintenance staff at Abitibi’s Fort Frances Mill with a valuable tool for doing control maintenance proactively and for detecting process problems unnoticed during routine maintenance. The use of a real-time asset monitoring and analysis system in conjunction with a control-tuning package has helped mill staff move toward proactive maintenance work processes to maximize the effectiveness of available resources while reducing variability, improving product quality, and preventing unplanned downtime.

Extensive process automation is a common feature of modern pulp and paper mills and a key component of competitive business strategies. As such, modern distributed control systems, along with associated regulatory process controls, constitute a significant part of a plant’s asset base. In addition to the initial investment, control infrastructure requires long term maintenance costs and management responsibilities if these assets are to continue providing business value.

Unfortunately, there are so many assets in a typical mill that a failure-based maintenance model has become the norm. As maintenance personnel attempt to correct pressing problems based on operator reports and scheduled preventative maintenance visits, underperforming control assets are often overlooked. In the current climate of rapid change, constrained resources and increased competition, plant management has focused on extracting more value from instrumentation and control investments to improve performance and profit margins.

Business situation

A global player in newsprint and uncoated groundwood, Abitibi-Consolidated’s pulp and paper mill in Fort Frances, Ontario produces 279,000 MT of value-added paper and 99,000 MT of NBSK market pulp per year.

A continuous improvement (CI) initiative provided the impetus for exploring new strategies to reduce variability, develop a proactive approach to maintenance and increase return on production assets. Some reasons the potential of automation assets was not being realized included:

• Lack of surveillance due to the large number of control assets
• Reactive maintenance allowed control loops and instrumentation to run to failure
• Off-quality product and unscheduled downtime due to failing control loops
• Lack of technical resources to effectively oversee control systems

Engineers proposed implementing a control asset monitoring solution on critical control loops in their Kraft mill. With this implementation, personnel hoped to move from a reactive maintenance methodology to a more proactive one by identifying and repairing poorly performing control loops before they negatively impacted plant profitability.

Mill staff had already achieved some positive results with a control-loop tuning package but it was limited by its inability to identify poorly performing control loops among the vast number in the plant, and the absence of a monitoring function to track how tuning changes affected overall plant control performance. Without monitoring and diagnostic tools to identify and prioritize control issues, opportunities to perform maintenance during planned shutdowns did not always address the most serious problems.

The desired solution, then, would allow personnel to continuously monitor a large number of control loops simultaneously. Further requirements included the ability of staff without process control backgrounds to understand and use the tool, and a non-intrusive implementation that did not require modifications to the DCS or plant historian.

Matrikon Control Performance Monitor was implemented on 350 control loops in the Kraft Mill and a number of loops in the ground wood mill. In total, the system was to provide continuous monitoring, diagnostic and tuning tools for just under 400 loops.

Matrikon Control Performance Monitor provides an automated means of assessing and monitoring the performance of all of assets in the control hierarchy up to and including the advanced control layer. According to Matrikon, the application is targeted at improving control performance while simultaneously lowering the long term cost of sustaining the performance of control assets. It is designed to integrate with existing plant maintenance processes, and is compatible with all commercial DCS and data historians.

Abitibi staff began by using the tools to identify loops that were cycling and to determine whether this was due to a malfunctioning control element, poor control tuning, or a disturbance-related problem. Maintenance efforts were directed to address the costliest problems first.

Reaping rewards

Six months after purchasing and applying Matrikon Control Performance Monitor in the Kraft mill, the process control team completed a study on the mill’s bleach plant, focusing on the 25 control loops around their R8 chlorine dioxide generator, where efficiency could be easily measured by comparing the quantities of input chemicals to the amount of chlorine dioxide produced. The study revealed three types of control performance problems and generated a quantifiable benefit in detecting and correcting these problems.

The primary controller for maintaining ClO2 solution strength was identified as a poor performer because:

• RPI (relative performance index) of 0.05. RPI values below 1 indicated a controller settling much slower than the desired benchmark. In this case the closed-loop settling time benchmark was set at four minutes and an actual settling time of 125 minutes was being achieved
• Closed-loop impulse response decayed much more slowly than the desired impulse response of a controller with a four minute settling time. Again, this sluggish behavior indicated an undertuned controller
• Closed-loop frequency response had much greater amplitude at the lower frequencies than the benchmark frequency response, indicating the controller to be significantly under tuned

The control-tuning package was used to identify new tuning parameters and Matrikon Control Performance Monitor was used to reassess and validate the control changes. After retuning, settling time decreased from 125 minutes to 10 minutes.

The absorption tower level controller also has significant impact on final product quality and was identified as showing room for improvement. Assessment with Matrikon Control Performance Monitor indicated a poor performance index, a large valve nonlinearity index and a cyclical response, symptomatic of a sticky control element. By correcting the failing device, control was improved, with a resulting increase in product quality and process efficiency.

The system helped engineers detect another control problem – with the R8 generator’s level controller that was impacting product quality and generation efficiency due to excessive chilled water addition, causing liquor concentration deviations from their optimum. Level in the R8 generator was regulated using an on-off control configuration. As level in the generator dropped below a desired set point, the make-up valve was opened fully until the setpoint was met, at which time the valve would fully close.

The control strategy was changed to a master/slave cascade arrangement with the generator level trimming the steam flow to the generator. The result was a dramatic reduction in chilled water use and liquor concentration deviations. In the four months prior to implementing the new control strategy, water was added to the generator at a rate of 140 hours per month. After the redesign the figure was reduced to 12 hours per month. This improvement reduced generator operating costs, simultaneously improving generator efficiency and product consistency and quality.

In the four months prior to the implementation of Matrikon Control Performance Monitor, the generator had an average operating efficiency of 88.7 percent. Fewer than 10 of the 25 loops audited were found to have problems; though some of these performed so poorly they had been switched over to manual control. In the four months following the implementation on the R8 Generator, the average efficiency rate rose to 93.4 percent, representing a cost savings of close to $400,000 annually.