Variability Management

Sometimes plants go through a revamp and things just don't work right. Loops that used to run fine in automatic mode now run in manual mode and require constant attention. I caught up last week with Emerson's James Beall, a principal process control consultant, whom you may recall from earlier posts.

James shared a great story with me about a recent visit he had to a refinery. Many months ago, they had gone through a crude unit revamp. They were struggling with the operations and it required quite a lot of operator intervention to keep things running. Many of the key crude unit loops were being run in manual because automatic mode was too unstable.

James described it to me as a back-to-basics visit. Working with the operations and plant engineering staff, they discovered quite a number of instrument and tuning problems. One example was a level transmitter that had been improperly configured. James used some of the control performance tools to gather and assess the process dynamics around the level loop.

While performing a bump test on the level loop, the team noticed the level signal responding in the wrong direction. A quick check revealed that the level transmitter calibration had set the transmitter to have reverse action, which prevented the level controller from working. It's no wonder the operators wisely had put the loop into manual mode to prevent this from happening.

Using a methodical one-by-one process through the critical loops, James helped the team properly tune each loop and discover problems with the surrounding control valves and instruments. All the tuning skills in the world can't help get a loop to behave properly if the valve is sticking or the transmitter is not performing well. The maintenance staff fixed what they could with their available spare parts and put a plan together for fixing the other problem assets.

Over the course of five days, the team resolved the major issues and smoothed out the operations of the crude unit. James shared with me that this approach helped the operations team regain confidence in the tuning and return the loops into automatic mode. His time spent with the team also was used to share his expertise to help them be able to work through future issues as they arise. In that sense, it was also in-the-field training on process control performance.

I like the quote he shared from one of the operators after the first two days on-site:

My goodness, you fixed more of these problems in two days than has been fixed in the past several months.

James noted that these results were really the result of a great teamwork effort with the plant personnel. Thanks for sharing this great story, James!

MP3 | iTunes

I've featured quite a number of experts around Emerson over these past three years. Some categories, like Process Optimization, have more than 50 posts. There is quite a bit of wisdom mixed in all those posts.

A few months ago, I was asked by Plant Engineering magazine managing editor, Jack Smith, if I'd be willing to write a process optimization article highlighting some of the ideas offered in these posts. It was a thrill to be asked and I quickly agreed.

I wrote a draft and went back to some of the experts highlighted, such as Mark Coughran and Pete Sharpe, for their improvement suggestions. The resulting article, Downturn a good time to review, improve process optimization went live on PlantEngineering.com website last Friday and is printed in the March 2009 magazine edition.

I tried to stress things you could do as a plant engineer to improve your process without having to spend a lot of capital, which is an issue for many process manufacturers in this global economic slowdown. Finding ways to reduce process variability is a good first step. Sources of variability that our variability consultants have tabulated over the years include:

• Control valve performance - 30%
• Improper tuning - 30%
• Improper process and/or control scheme design - 20%
• Other - 20%. The 20% of other causes are not necessarily design- or control-scheme related, but more operational issues that occur over time.

I distilled down five ways to reduce this variability: size control valves properly, minimize loop dead time, measure process dynamics and compensate for them, tune the loops, and apply advanced process control. I won't spill all the secrets divulged in the article but instead highlight a couple of points.

My Fisher valve colleagues often remind me of the importance of the control valve since it directly touches the process.

Control valves, being variable in gain, must be correctly sized and characterized for the application's flow to be sufficiently linear to stay within specified gain limits over the operating range of the process.

Other parts of your control loops to check for:

...sources of dead time include inadequate signal conditioning on transmitters, incorrect transmitter range/resolution, poor physical location of transmitters and measurement lags from applied filters and dampeners.

Without the proper process dynamic measurement applications, many plant engineers have had to rely on rules of thumb and guesswork to loop tuning parameters. With the process dynamics understood, you can tune the loops with linear responses and try to reduce the non-linearities in the others in several ways including:

...changing any master loop configuration to prevent interaction with the slave loops. See if you can remove unnecessary interlocks that may disturb the control loop. If you uncover extremely high process gains, adding upstream control loops can help. Other advanced regulatory control strategies such as feedforward, cascade, override and split-range control can compensate for different process conditions.

With the basics addressed, you can look for areas to apply advanced process control, especially in big, energy-consuming units.

I hope some of the ideas excerpted from the article help you find ways to improve your process and help your business through these challenging times.

MP3 | iTunes

Update:I heard from Jack that this article is currently the most popular one on the PlantEngineering.com website. Thanks for stopping by to read it!

If ever there were a need for dynamic control, it would be for the global bond and equity markets this week. Instead of tackling that global challenge, with today's post I'll highlight Emerson senior process control consultant Mark Coughran's Emerson Exchange presentation. He presented, "Equipment Selection and Installation for Dynamic Loop Requirements" last week.

The purpose of Mark's presentation was to share ways to extract more value from the control system by designing it for the process dynamics or to diagnose and improve control in an existing installation by better understanding the process dynamics.

A key dynamic control requirement for the loops is that process variable (PV) changes should move to a new setpoint (SP) without oscillation. Recovery from process disturbances should also occur without oscillation. The loops should run in a cascade or automatic mode without the need for operator intervention. And finally, the loops need to respond at the speed required to meet the process objectives.

Mark stressed the importance of properly selecting, installing and maintaining the final control elements (like control valves), transmitters and controllers. No amount of tuning can overcome improperly sized or malfunctioning loop elements.

Given Mark's experience in working with process manufacturers, here's his top-ten list of mistakes that he encounters:

• Oversized control valves
• Inappropriate valve flow characteristics
• Non-communicating, low-performance valve positioners
• Positioners not tuned for the damper/actuator
• Inadequate signal conditioning (thermocouples)
• Incorrect range/resolution of transmitters
• Poor physical location of transmitters
• Oscillatory tuning of loop controllers
• Unit controller tuning ignores interactions
• Inappropriate control logic for switching block valves

Mark's presentation included a number of specific examples including refinery boiler feedwater control, reactor jacket temperature control, Olefins plant reactor temperature control, boiler O₂ control, acetylene reactor temperature control, beverage-quality CO₂ heat exchanger gas pressure control, pulp mill lime mud density control, Lysine plant resin bed level control, and chiller level control. This is a wide-ranging list of control challenges, but the common thread is to start with an understanding of the process dynamics by measuring them.

The solutions Mark discussed ranged from control valve trim changes, to Lambda loop tuning, to thorough operating training, to application of advanced control strategies like model predictive control.

Some of the common threads of Mark's guidance were:

1. Use high-performance smart positioners
2. Specify the control valve/damper dead band, response time (ANSI/ISA-75.25.01-2000)
3. Select the valve size and flow characteristic for installed gain
4. Use a signal characterizer or PID gain scheduling if process gain varies greatly
5. Tune the positioner for the valve/actuator
6. Install the sensor for minimum dead time
7. Use a transmitter to amplify low-level signals
8. Select the appropriate transmitter hardware span
9. Apply Lambda tuning for non-oscillatory response
10. Apply Lambda tuning to coordinate response of interacting loops
11. Check other control logic outside the PID block

None of this might help with the roiling financial markets, but it can certainly help your process run more smoothly and efficiently.

MP3 | iTunes

I caught up with Emerson's Mark Coughran, a senior process variability consultant whom you may recall from earlier process tuning and optimization posts.

Mark shared a story of a plastics manufacturer that was challenged to bring a new product to market with a new extruder. This manufacturer needed to run trials with varying polymer formulations at various temperatures and speeds while trying to perfect the production process.

The plant control engineer was struggling with control strategy and necessary tuning to hold the required temperature. The temperature loop wasn't responding to setpoint step changes and was oscillating even when no disturbances were present.

When Mark arrived to lend his experience to this challenge, many anxious folks greeted him. The project engineer was glad to see him. The project manager asked if he could stay the weekend. The plant manager assured Mark that the situation had visibility at the highest levels of the organization. A corporate engineer added pressure by saying single loop control worked just fine at a similar plant. I imagine that Mark didn't enjoy all this attention.

He and the project engineer began by measuring the process dynamics--both the linear and non-linear components. For the linear process responses, Mark applied Lambda tuning.

For the non-linear portions of the overall process dynamics, the approach was to mitigate these nonlinearities as much as possible. They performed four actions to accomplish this. The first step was to improve the control strategy by changing the master loop configuration to prevent interaction with the slave loops. Next Mark helped identify and have an unnecessary interlock removed that disturbed the control loop.

The process dynamic measurements uncovered an extremely high process gain, which was reduced by establishing pressure control upstream from the extruder. Finally, the output pulsing was adjusted to better match the control strategy with the control valve dynamics.

After applying these changes, the temperature process variables tracked the setpoint changes over the operating range of the trials. Mark typically likes to work with the process manufacturer to financially quantify the results to prove the value of his services. It also helps the people he works with look good to their upper management. Unfortunately, until this new product gets to market its value is not yet determined and the control engineer didn't want to speculate. In this case, robust control was established and the level of anxiety dropped considerably.

I was reading the great Control magazine article, Data Analytics in Batch Operations, by Lubrizol's Robert Wojewodka and Emerson's Terry Blevins. It describes the challenges in applying on-line analytics in batch production for early fault detection and prediction of end of batch quality parameters. If done right, this can help avoid out of spec batches, waste and rework and the opportunity cost of the lost batch. This can be critically important in specialty chemical and pharmaceutical manufacturing, which in the words of the authors, "depend heavily on batch processing to produce low-volume, high-value product."

The reward is great if you can make these on-line analytics work but the challenges are great. One of the biggest is the time differences between batches. Operators and events can halts and restarts to process. These may be for manual additive additions, waiting on common equipment to become available or abnormal conditions that may develop.

Another challenge is that online measurement of quality parameters "may not be technically feasible or economically justified." These lab samples must be available to online analytics toolset to perform these analytics.

Other challenges they noted included variations in feedstocks, varying operating conditions, concurrent batches and the assembly and organization of the data.

The concept of "Golden Batch" has been around a long time, which is the concept of comparing batches in progress, or just completed with ideal ones from the past. The authors point out two big weaknesses with this approach. The first is conditions indicated by each measurement may affect the product quality in different ways. Secondly, this is a univariate approach to a multivariate problem--no knowledge is gained of the relationships of process variations.

In a prior post, I've discussed some of the analytic tools, PCA and PLS. By applying these online, "...changes can be made in the batch to correct for detected faults or deviations in the predicted value of key quality parameters."

I mentioned to Terry that I gave this article a close read and was working on a blog post. He told me the really innovative thing they were able to do was to apply dynamic time warping (DTW). The article describes its use:

...allows such [batch time length] variations to be addressed by synchronizing batch data automatically using key characteristics of a reference trajectory.

This normalization process is covered in detail in the Greg McMillan and Mike Boudreau's book, New Directions in Bioprocess Modeling and Control and discussed in the article, PAT Tools for Accelerated Development and Improvement.

It's hard to give the article justice in a short blog post, so if you have 18 minutes and 30 seconds to spare, watch to the interview Walt Boyes did with Bob, Terry and Philippe Moro at last year's Emerson Exchange.

James Beall delivered a Back to the Basics - Process Control Diagnostics Improves Refinery Performance presentation at the recent AIChE spring meeting. James, whom you may recall from earlier variability management posts, is a principal process control consultant. He's a senior member of Emerson's variability management consulting team.

In this presentation, James stressed what he normally stresses with process manufacturers--that some of the largest and most frequent opportunities exist in basic process control. These opportunities include eliminating variability at the source, tuning the controllers to meet the control objective, using ratio, cascade and feed forward control as well as using a process analysis system to diagnose problems and tune loops. Addressing these opportunities also builds a control foundation essential for any advanced process control (APC) initiatives.

He referenced a 1997 McKinsey study that showed 50-60% of the value realized from a process optimization project comes from addressing loop variability. The balance 40-50% comes from applying APC on top of these optimized loops. The financial results from reducing variability are being able to operate closer to constraints such as specification limits. Benefits can come from reduced energy consumption, less waste and rework, higher yields, higher quality, etc.

The variability management team keeps statistics on control loops with excessive variability from site audits. The major causes of this variability include control valve performance (30%), improper tuning (30%) and improper process design (20%).

James shared several valve-performance examples including a regenerator pressure valve. By looking at the setpoint, pressure, output, and valve position trends, he spotted the valve sticking and then jumping 3% followed by a quick spike of another 2-3%. This caused periods of oscillations before settling out. Once the sticking problem was addressed, the oscillations became tiny ripples on the trends. Similarly, poorly tuned loops can cause large oscillations impacting overall process variability.

He noted that you must have a process dynamics analysis and diagnostic tool of some type to pinpoint these sources of variability. Problem identification is the first step in corrective action. And these problems may be significantly impacting the overall efficiency of the process.

James described some of the tests that he and the variability management consultants use with the Entech Toolkit. One of the most important tests is to identify the process dynamics so that the control loops can be properly tuned. Emerson's Entech Toolkit can identify common dynamics such as first order, second order overdamped and integrator+lag. Dynamics that are more complex can be identified by this process analysis toolkit (11 tests in total) and the associated controller can be properly tuned. Many of the more complex process dynamic responses cannot be identified by less sophisticated analysis systems.

If you have the bandwidth and inclination to learn the skills to do it yourself, James recommends three Emerson Education Center courses: Process Dynamics, Control and Tuning Fundamentals, Process Analysis and Minimizing Variability and Modern Loop Tuning.

In the upcoming March issue of BioProcess International magazine, there is a great article by Emerson's Greg McMillan and Michael Boudreau, Broadley-James' Trish Benton, and the University of Texas at Austin's Yang Zang. The article, PAT Tools for Accelerated Process Development and Improvement, describes the collaborative effort between Emerson, Broadley-James, and UT, "...to examine and quantify the potential for faster optimization of batch operating points, process design, and cycle times." The specific objective of this collaboration:

...is to show that the impact of PAT can be maximized through the integration of dynamic simulation and multivariate analytics in a laboratory-optimized control system during product development.

Greg and Michael are putting many of the ideas they described in their book, New Directions in Bioprocess Modeling and Control: Maximizing Process Analytical Technology Benefits, into practice.

The authors outline the challenge for the 400 biotechnology medicines currently in development, which require overlapping and iterative stages for process development and commercialization. These stages include:

...cell line selection and development, media optimization, process conditions optimization and verification, scale-up, project definition, and plant design.

This team is working on beta tests using this new dynamic model and on-line data analytics and wants to make the results fully public to promote wide use and to advance these concepts and methodologies.

If you're like me and not in the biotechnology field, much of the article may get a little deep. I did glean a few tidbits you might find useful. By creating a dynamic model, one of the big benefits to the team is the ability to speed up the model by up to 1000 times real-time. Whether you're simulating the growth of mammalian cell lines or have another slow process, this can really help reduce trial and error time.

Another key is that the model, configuration, and tools can run in the "virtual plant" PC environment or can be downloaded to the automation system. With proper scale up factors:

...the embedded tools go readily from bench-top bioreactors to pilot plants and eventually industrial-scale bioreactors.

With the recognition by the FDA that quality cannot be tested into products, which led to the creation of the Process Analytical Technology (PAT) initiative, the authors discuss the role of analytics in their efforts.

Principal component analysis (PCA) and projection to latent structures (PLS) are two multivariate analysis techniques that can help analyze continuous and batch process operations. The authors' beta test is focusing on the on-line use of these analytical techniques where PLS detects deviations in quality parameters and PCA detects abnormal operations from measured and unmeasured disturbances.

Given the importance of new product development for pharmaceutical and biotechnology manufacturers, anything to reduce the overall development time and build in quality monitoring as prescribed in PAT should be a welcome addition.

The Automation List on Control.com recently had a question about IMC Tuning for Integrated Processes. I googled around for IMC or internal model control for a good definition and found these 2002 Introduction to Robust Control lecture notes:

The Internal Model Control (IMC) philosophy relies on the Internal Model Principle, which states that control can be achieved only if the control system encapsulates, either implicitly or explicitly, some representation of the process to be controlled.

The Automation List question asked how IMC can be implemented if the process time constant, process gain, control integral and controller gain are unknown. This person did a manual step test on the drum level feedwater control valve and the drum level starts to integrate (rise.) Measuring this occurrence provides dead time, level rate of change and change in control valve position.

The questioner writes:

I know you can implement Lambda Tuning, but from what I've seen with this, you end up with a very sluggish system that responds quite poorly due to the low value of Kc (please don't comment here on 3 element control, as this is not apart of the discussion).

Am I missing something here, or have other people used different methods?

We've had several posts in the past on Lambda tuning, so I forwarded this question to Mark Coughran, a senior control engineering consultant on the Advanced Applied Technology team.

Mark notes:

Whatever method you use, it is important to understand each of the terms in the equations and the appropriate units of measure. Training is available to make clear how to measure the process dynamics, choose Lambda, and calculate the controller gain and reset. Emerson Educational Services offers the courses Process Dynamics, Control and Tuning Fundamentals (9030) and Modern Loop Tuning (9032). Tools and on-site services are also available.

Lambda tuning simply means the loop will not oscillate and you choose the speed of closed-loop response (Lambda), within some reasonable constraints. There is no reason to believe that Lambda tuning is arbitrarily "slow" or "fast", since you choose the Lambda.

ZN or Ziegler-Nichols is a method to deliberately make the loop oscillate. This is not a good idea in any process plant.

In spite of my best efforts to use persistent RSS search feeds in order not to miss any news about Emerson experts in action, here's one that got by me.

Mark Coughran, a consultant on Emerson's Advanced Applied Technology team, shared this control challenge question he answered with me. You may recall Mark from earlier posts.

The question he addressed appeared on the ChemicalProcess.com's Ask The Experts website. The question, Control pressure at discharge, was:

I have five pumps running parallel, transferring water. Due to pressure fluctuation at discharge, which depends on the flow requirements of the user, I am planning to install a pressure control valve at the pump discharge to keep the pump running at an optimum condition... What kind of valve is best for a 14-in discharge?

Mark notes that he's seen problems with butterfly valves used on large water lines, but that things have improved with better valve, actuator, positioner, and application software. Common sources of problems include wrong valve size, shape of butterfly disk, backlash in disk-to-shaft and shaft-to-actuator connections, poor valve positioner performance, and insufficient torque.

Control valve suppliers have addressed these issues in a number of ways. Examples include better valve sizing software, improved butterfly valve disk shapes, zero-backlash connections, valve positioners responding to 0.1% signal changes, and sizing software that predicts installed torque.

Mark points out that globe valves are typically too expensive for this application. Butterfly or segmented ball valves may be better suited if the supplier's test data for the valve + actuator + positioner shows suitability in similar applications.

Mark's final guidance concerns the control strategy. He recommends a controller tuning method that does not oscillate, but responds at the application's required speed, such as Lambda tuning. He advises:

If you need to control the five lines separately, there will be interaction and balancing concerns. The options range from individual PID controllers to a multivariable controller. All the options are easy to configure and tune in a modern DCS.

The Automation.com list server has an interesting thread, Three Element Drum Level Control Problem. The question asked was:

We have a waste heat recovery boiler that is supplied by exhaust of a 20MW Gas Turbine. We've seen that at lower turbine loads (75% and below) the three element drum level controller cannot maintain the drum level at desired setpoint. As soon as the load on the Gas Turbine is increased to more than 75% of rated load, the stability keeps getting better. At rated load (20MW) the drum level is very stable and close to the setpoint.

There have been several responses discussing the tuning at various loads. I asked around to see what advice we might have to offer. Emerson's Jack Tippett, a variability management consultant noted that it is critical to know your process dynamics. His point:

If you don't know the process dynamics, control tuning is an art not a science and good control performance is an accident not a certainty.

Once you know your process dynamics, it is important to design your strategy to assist in achieving the process objectives in light of those dynamics. Jack noted a similar situation from his past where he tuned the levels in a 450-megawatt heat recovery steam generator (HRSG) system.

There were six boilers including two lines with high, medium and low-pressure drums. This power producer was unable to achieve a station ramp rate of 25 MW per minute necessary for automatic generation control (AGC) due to serious swings in the drum levels.

After measuring and determining the process dynamics, the process was re-tuned and they were able to achieve the ramp rate and achieve good level control at less than 70% load.

Jack also noted that they chose a single-element control strategy for the following reasons:

1. Feedwater flow control requires a working flow meter: the sense lines for the flow transmitter were outside and were subject to freezing. The Fisher valve had a DVC positioner and AMS software to monitor incipient valve non-linearities (which are the main reason for the second element.)
2. The open loop dynamics (changing the feedwater valve position manually and watching the response to level) on all six boilers showed very small dead times (1 to 6 seconds). This meant that the proportional-integral (PI) level tuning could be very aggressive. As a result, there was no value in the third element (steam flow feed forward)--the level control could be fast enough to respond the changes in level due to steam demand changes. The real need for the feed forward from steam is when the level dynamics are very slow (30 - 90 seconds dead time) so that the feedwater flow can anticipate the long-term level changes (due to steam demand) in spite of the shrink/swell effect.

By having good measurement in the flow, valve position, and valve characteristics and good understanding of the process dynamics across its operating range, Jack and the plant engineers were able to successfully implement a simple single-element control strategy.

Like capacitors do for electrical circuits, vessel levels provide capacity that can absorb variability within the process. In many cases, a properly tuned level controller can make the variability of the vessel outflow can be much less than the variability vessel inflow. Of course, if it's not properly tuned, the variability can pass right through the vessel or even be amplified. In fact, an improperly tuned level controller can make the variability of the outflow higher than that of the inflow! Unfortunately, the latter cases are common and directly impact process efficiency and product quality.

I caught up with Emerson's James Beall, whom you may recall from earlier posts. James is a process control veteran with 27 years of experience including the last seven as a process control consultant. He's also chairman of the ISA 75.25 committee on control valve dynamic testing.

James stressed that there are often different objectives in tuning level controllers. What's common is to make sure this tuning is not creating variability on its own. Sometimes you want to hold the level very close to the setpoint at the expense of aggressively moving a manipulated variable. Other times, like in the case described here, you want to use the capacity of the level system to absorb variability in the process and very smoothly move the manipulated variable as little as necessary.

From James' experience, good level tuning techniques for absorbing process variability are not widely known. When the level controller for a vessel is properly tuned, variability can be reduced by a ratio of a 20:1 or more depending of the nature of the variability and process constraints. For example, a vessel with an input flow which varies plus or minus 20% of the inflow can have the variability reduced to plus or minus 1% of outflow.

Most level processes have an integrating process response. This means the level is an integration (or accumulation) of the difference between the inflows and outflows of the vessel. Absorbing process variability requires that the level control be tuned as slow as possible but still fast enough to hold the level within the allowable deviation for the maximum expected load change.

For an integrating process, Lambda is the period of time from the start of the step change in load until the process variable has stopped changing as a result of the level controller action. The level controller tuning parameters can be calculated to achieve a specific Lambda.

James notes that the required Lambda for maximum variability reduction is a function of the allowable level variation, the product of the integrating process gain and the maximum expected load disturbance. He uses this technique on a number of applications including feed tanks, distillation column bases, intermediate tanks, and reflux accumulators.

He mentioned that in some cases, the level actually needs to be controlled very close to the setpoint. Examples of this case include boiler drum levels, refrigeration evaporator levels, reactor levels, etc. He also has techniques for calculating the level controller tuning for "tight level control" which we'll explore in future posts.

James has worked with process manufacturers where reoccurring annual benefits from proper loop tuning have yielded savings from several hundred thousand dollars to several million dollars.

If you have the time and inclination to learn more about these concepts, James recommends three courses offered by Emerson Educational Services: Process Dynamics, Control and Tuning Fundamentals; Process Analysis and Minimizing Variability; Modern Loop Tuning.

If you don't have the time or inclination for these courses but need help in addressing process variability issues like level control, you'll want to connect with James and the variability management team.

Imagine that your power plant is about to have a scheduled outage. As the unit is ramped down and feed water control is taken over by the by-pass valves, you discover that the control valves refuse to close upon receiving orders from the level control system to do so. This is now the last straw for the operators who also have been fighting stability problems with these valves over the past several years. What do you do?

Well, if you know Emerson variability consultant, Eric Ascoli, you contact him. You may recall Eric from a prior post on stability problems at a sugar mill. He shared this story with me.

Instead of continuing with the shutdown, the station had to run at 20% power production for 12 hours costing them hundreds of thousands of dollars while the situation was diagnosed and corrected. A manual unit trip was not an option. The problem was aging pneumatic instrumentation that had locked up and blocked the valve positioner's operation.

Eric worked with a pneumatic specialist from Proconex, the local business partner for the power station. Their findings were that the operation of the pneumatic trip valve was not completely understood and its adjustment was slightly off. Also, the combined level control valves had a very large variation in installed valve gain and the unbalanced and aggressive controller tuning caused the instability the operators had been experiencing.

The challenge was to find a solution that would remove completely the possibility of such an event from happening again. It involved a short-term fix (servicing and adjustment of the pneumatics and modification of the characterizing functions for the valves) in preparation for the imminent scheduled power up. Additionally, Eric corrected the level controller tuning by using Lambda tuning after he analyzed and evaluated the process gain and empirically defined other important process parameters.

Their longer-term recommendations were to install digital valve positioners to replace the aging pneumatic ones. The same split-range control strategy would be maintained, but the valve performance would be improved through better positioning accuracy and dynamic behavior. The installation would be simpler and less prone to maintenance issues because I/P (current to pneumatic) converters could be removed. An even longer-term solution would be to replace the two split-ranged valves with one single start-up control valve to eliminate any crossover interactions.

I've highlighted the topic of plant turnarounds (planned downtime for maintenance) a few times in the past. Back from the Emerson Exchange, here's my take on the Smart Turnaround workshop. For continuous processes that run for years, this turnaround provides opportunity to update, fix, repair, and replace a host of plant assets including instruments, valves, electrical distribution equipment, connectors and cabling, and the overall performance of the process.

The Emerson presenters looked at the advanced planning that can be done from these various perspectives. From these diverse areas of expertise, diagnostic testing helps develop a turnaround plan that prioritizes critical asset work, defines the scope of work, develops the schedule for the work, and identifies the parts and people required to best get this difficult work done.

Chris Forland an operations consultant whose work I've highlighted in earlier posts kicked off the session discussing some of the challenges of the turnaround process. A big one is finding problems you didn't expect while in the turnaround. These unexpected problems cause extra charges and delays. Chris discussed ways that Emerson turnaround specialists can help with the detailed planning to make sure the work is efficiently performed during the turnaround. He noted that less time to plan mean less flexibility as the turnaround date approaches. Other challenges included maintaining compliance with safety and regulatory compliance, working with budget constraints, reducing process variability, losing experienced personnel due to infrequency of turnarounds, and pressuring of short turnarounds due to sold out condition of produced product.

Scott Grunwald, a turnaround business manager in the Instrument & Valve Services business, recommended that with the valves and instruments, you start by building the plan based on the benefits to be achieved the roles of all participants in the maintenance activities, and the prioritized list of activities and anticipated timelines. The process starts with a walk down of the facility. Next, FlowScanner is used to measure internal valve conditions to identify problems to address during the turnaround. When it's time for executing the turnaround, only valves needing significant work are removed. Other valves are repaired in place.

The team often brings an on-sight mobile trailer that is a self-contained workshop to rework the instrument and valves right on-site. This helps to expedite the repair process.

Looking at turnarounds from an electrical reliability perspective, Steve Metzger described the goal--to prioritize and focus the resources by pre-diagnosing troubleshooting, followed by the planning of the repair services and parts required to get the lead times properly. The key is to do as much pre-work as possible, fix what's possible, and remove it from the scope of the turnaround to lessen the pile of work to be done.

On-line partial discharge testing before the turnaround detects cables with degrading insulation that could cause short circuits and unexpected downtime. This testing helps determine which cables are OK and which need to be replaced during the turnaround.

James Beall, also highlighted in earlier posts, summed up the goal of a Smart Turnaround--to identify the items you can fix in advance, and prioritize what can't be in the turnaround plan. James and the variability management consultants look at the control performance and opportunities to reduce process variability through better tuning. James gave an example of a mixing temperature control loop where the deadtime was nine minutes between a change in setpoint and response the temperature was changing. The problem was not in the loop tuning but rather in the lag caused by the temperature transmitter being located 250 feet from where it should have been. Finding this early in the process allowed this installation mistake to be scheduled and fixed during the turnaround.

Chris closed this presentation with how you can look at the return on investment to help justify the experts required to make the planning and execution of the turnaround a success. It's a bit of a chicken and egg scenario since you don't know what type of ROI this turnaround planning can create without having the experts come in to begin the process of identifying improvement opportunities.

Chris has developed a model based on turnaround experience with typical costs from each of the aspects of turnaround planning and typical costs for the maintenance activities. This model is in an excel spreadsheets so that the assumptions can be easily changed to fit the unique aspects of each process manufacturer. Both cost avoidance and increased revenue from improved plant performance is calculated, each based on the size of the process and amount of equipment considered.

By taking a comprehensive planning approach, and getting an early start, turnarounds do not have to cause quite the number of gray hairs that they have traditionally been known to cause.

Update: Mitzi Amon, director of marketing for Emerson Electrical Reliability Services team adds that the prioritization is accomplished by performing online diagnostic testing prior to the turnaround to determine what electrical equipment needs to be serviced during the turnaround. This helps clearly define maintenance work scope during the turnaround and what can be done prior to the the turnaround.

Recently, a North American chemical manufacturer was having problems with their boilers tripping during startup and shutdown sequences. This problem was caused by a wide variation in the process' demand for steam. This situation caused lost production, which affected the overall plant efficiency.

Jim Dunbar, an Emerson variability management consultant was called in to provide emergency tuning services, to set the loops on the boilers to be able to handle the range in steam demand.

Jim's mission was to work with 2 boilers and about 10 loops controlling these boilers to resolve the situation.

The problem began when the plant installed a new steam-driven compressor that required a minimum steam pressure for operation. The team installed a backpressure controller to satisfy the steam requirements of the compressor. However, the boiler still had to ramp up very quickly to maintain the plant steam header pressure on process unit shutdowns. When the boiler firing-rate was increased too rapidly, the boiler would trip due to low feedwater level.

Jim worked with the plant staff to perform open loop bump tests on the feedwater flow and drum-level control loops. This data was collected in the PI historian and analyzed with the EnTech Tuner. Lambda tuning constants were calculated resulting in much faster and stable drum level control. Next, the boiler master controls were tuned to coordinate the speed of response with the level control. It was important that the firing response was fast enough to meet the requirements of the steam header, but not so fast as to cause an unrecoverable upset to the drum level resulting in a boiler trip.

Since his visit to the site, the manufacturer has not had a boiler trip in over four weeks, despite numerous simultaneous unit shutdowns.

Beyond the improved reliability of the process, Jim provided the operations staff some key insights on what to watch for if instability creeps back into the process.

In an earlier post, I mentioned seeing a draft article by Emerson's Terry Blevins and James Beall on performance monitoring and the Process Analytical Technologies (PAT) initiative.

The article, Monitoring and Control Tools for Implementing PAT, has now been published in Pharmaceutical Technology magazine.

Terry and James do a great job in summarizing the common problems process monitoring can detect. These include problems where the control is limited, information from the field transmitters is bad or uncertain, loop modes are incorrect, or there is high variability associate with the loop.

You can do process monitoring with an application that runs either on top of the existing automation system or embedded within it as I discussed in an earlier post on DeltaV InSight.

Here's a few tips gleaned from the article which I'll paraphrase:

• Make sure the performance monitoring application understands the operating states of the batch process avoid false indications or failed measurements
• Where you are using smart field devices like Foundation fieldbus, HART, or others include the status which accompanies the measurement so that performance calculations are based on valid information
• Check the operating modes of the loops versus their design as a basic measurement of control performance.
• Having a model to compare the actual running process against can help spot the largest areas of variability to focus improvement efforts.

Terry and James wrap up their article nicely pointing out that the Food and Drug Administration's PAT initiative has opened up the opportunity to use these performance monitoring tools to improve the operations of their processes. The timing is great with newer technologies coming along to simplify the performance monitoring process.

It is great to see the area of Emerson Process Management website on variability management services updated. I received a heads up from Bill Tjoa who manages the marketing efforts for this and other consulting programs.

On the right-side navigation, the site links to the variability management posts we've done here at Emerson Process Experts. With the addition of these posts, the pages provide a nice blend of services, stories and whitepapers of the people who perform these services, seminar topics/dates/locations, and tools like the Emerson Entech toolkit.

Over the past couple of years, more and more resources have become available to help you better understand ways to deal with process variability.

Over at the ModelingAndControl.com blog, Greg and Terry share the ins and outs in controlling a process given deadtimes, disturbances, non-linearities, and all the other challenges automation engineers face. You can learn from their practical experiences and apply it in your manufacturing processes.

In the educational services area, I've been pointing to John Egnew's Loop Tips on practical tips to improve loop performance.

Over here, we've shown a screencast how newer technologies automatically learn the dynamics of your process.

And with the variability management consultants, you have experienced folks who can help apply their expertise to help you reduce variability. Better controlling this variability can yield operational improvements and quantified business results like those chronicled in an earlier post:

The columns achieved 100% of design production 5 days after James first arrived and nearly another 20% after a total of 7 days.

We'll continue to look for opportunities to share their stories here.

I came across the following nine control fundamentals according to Mark Coughran, a consultant on Emerson's Advanced Applied Technology team. These are based up his years of experience working with process manufacturers to optimize their performance. You may recall Mark from earlier posts on planning plant turnarounds and turbomachinery pressure control.

His fundamentals include:

1. Make the process as linear as possible
2. Minimize dead time
3. Choose the PID controller to compensate for the process
4. Avoid resonance or amplification of disturbances
5. Use process capacity to absorb variability
6. Decouple the interactions by tuning if possible
7. Help the PID feedback controller with control strategy
8. Cascade, ratio, feedforward; a.k.a. advanced regulatory controls
9. Use Fuzzy, Neural, MPC if the above are insufficient

Although there is a lot behind each one, it's a way to think through the process of solving control performance issues.

Mark cited an example of a Good Automated Manufacturing Practice (GAMP) facility with eight reactors. All of the controls cycled strongly. As a result, steam was being wasted on the up cycle, and cooling water was being wasted on the down cycle. Thinking through the control fundamentals above, Mark recommended changes in the master controller parameters including tuning, jacket controller tuning, split range strategy, and control valve calibrations.

By implementing these changes on three of the eight reactors steam usage for the facility was reduced 10% reducing the plant's energy bills.

I caught a sneak preview of draft article that ModelingAndControl.com's Terry Blevins who collaborated with James Beall whom you may recall from earlier posts.

The draft explores the initial steps Pharmaceutical and Biotech manufacturers should consider when preparing to implement the U.S. Food and Drug Administration's Process Analytical Technology (PAT) initiative. For those unfamiliar with the PAT guidelines, they were established to encourage innovation in development and implementation of manufacturing processes to improve product quality. The existing regulation designed to achieve quality through rigorous design and documentation actually served to discourage improvements due to the time-consuming nature of the revalidation of any changes.

Terry and James offer some guidance on some initial steps that Life Sciences manufacturers can take. Since most of their manufacturing processes are batch-based, it can be trickier to apply some of the advanced process control technologies more often found in continuous processes found in the chemical, petrochemical, and oil and gas industries. They recommend starting by looking at ongoing performance monitoring. This software has typically layered on top of the automation systems but has begun to become embedded in the automation system. DeltaV Insight is a good example of this type of performance monitoring software embedded in the DeltaV system. This performance monitoring can be keyed to the phases within the running batch to account for the changing process conditions. The dynamics of the process are learned as changes in the process are made.

Terry points out that these performance monitoring tools can help manufacturers spot issues like excessive process variability which can have direct impact on product quality. Other conditions this software can help detect include control-limited conditions, bad/unreliable data coming from intelligent field devices, and control loops operating in modes other than those intended. All of these conditions can contribute to quality issues in the final product.

He notes that the ability of intelligent field devices to provide status of the goodness of the data is a key part of performance monitoring so that the control strategies, history collection, and analytical tools have a clear picture of what is really happening in the process.

I look forward to seeing the finished article!

I saw an email about a success at a northern U.S. paper maker that set a new production record. To whom did they attribute this success? Since this is a blog about the experts around Emerson Process Management, you might guess the answer. And you'd be correct. They attributed their new production record to the work of our Control Performance team and their process and control study process.

I caught up with Andrew Waite, a principal process control consultant on the Control Performance team. Andrew began the study by using the EnTech toolkit which collects data from a variety of sources including pneumatic controllers, 4-20mA analog values, and can import digital data from smart field devices and digital automation systems using the OSIsoft PI data historian. The toolkit performs analysis and tuning recommendations based upon the data it collects.

Andrew noted that he uncovered all of the typical problems: tun