Continue reading »]]> The ISA104 Device Integration committee was established to develop or adopt, and to facilitate the application of standards and technical reports that represent device and interface descriptions for access of field devices in automation systems. In 2012 the scope of the committee was expanded to include the promotion of the Field Device Integration (FDI) standard, IEC 62769. The FDI standard integrates concepts of both the Electronic Device Description Language (EDDL, standardized in IEC 61804-3) and Field Device Technology (FDT, standardized in IEC 62453). This new technology reduces efforts for device manufacturers since FDI standardizes a common information model for field devices. To communicate and promote the benefits of the FDI technology, four members of the ISA104 committee have written a paper, “FDI Meets Plant’s Device Integration Needs” that will be presented on the first day of ISA Automation Week. This year ISA Automation Week will be held 5-7 November 2013 at the Nashville Convention Center, Nashville, TN.
In this paper we address the challenges that users of 4-20 mA/HART, FOUNDATION fieldbus, and PROFIBUS devices currently face with two sets of files to integrate their devices; EDD files and DTM1.2.x software drivers. Information is provided on how The Field Device Integration technology better meets plant needs, which includes investment protection, robustness, easy system administration, easy to use devices, interoperability, and easy migration.

The FDI technology combines the best features of EDDL and FDT2 and provides a single device package that can be used on any system. The paper discusses how FDI meets the recommendations released by NAMUR chemical industry user organization, some of which are contained in their NE105 recommendation.

If you would like to learn more about FDI technology then I encourage you to sit in on this paper presentation at ISA Automation Week. Also, visitors are welcome to sit in on the ISA104 committee meeting that will be held on Monday, November 4th at the Nashville Convention Center – one day before the conference opens.

Continue reading »]]> A paper published in 1960 by Rudolf Kálmán “A New Approach to Linear Filtering and Prediction Problems” is the basis for the Kalman Filter. The Kalman filter has been successfully used in a wide variety of applications:

• The guidance of commercial airplanes
• Seismic data processing
• Nuclear power plant instrumentation
• Vehicle navigation and control (e.g. the Apollo vehicle),
• Radar tracking algorithms for ABM applications

However, the complexity of the Kalman filter algorithm is often a barrier in the application of this filtering technique in the process industry. The original Kalman filter as illustrated below was designed to address a general multivariate environment where the process and measurement noise covariance Q and R are known and used to dynamically calculate the Kalman gain, K.

At Emerson Exchange, 2013, Willy Wojsznis and I will host a workshop (8-4361) “Addressing Control in the Presence of Process and Measurement Noise” in which we will present a practical application of a scalar Kalman Filter. In the workshop we discuss the expected improvements in closed loop control that may be achieve when the control measurement is characterized by significant measurement or process noise. Also, we will show a DeltaV linked composite available through the Application Exchange (in late July, 2013) for implementation of the Kalman filter. This composite is designed for use with the PID function block in closed loop control and may be installed on any version of DeltaV. Information will be provided on a DeltaV module that may be used to demonstrate and get more familiar with the Kalman filter in a test environment.

If you are interested in learning more about how the Kalman filter may be applied in DeltaV, then I encourage you to attend this workshop at Emerson Exchange 2013. In case you are not able to attend Emerson Exchange, I will post the workshop presentation in a blog on the workshop after Emerson Exchange.

Continue reading »]]> Before tuning calculations are applied in a control loop, it is good to consider how much the process gain and deadtime can change before the loop becomes unstable. In the DeltaV control system, this information is provided using a robustness plot that shows gain margin and phase margin as illustrated below.

Information on this type of robustness plot is provided in Chapter 4 of Advanced Control Foundation – Tools, Techniques, and Applications. In this chapter, gain and phase margin are described in the following manner:

• Gain margin simply means how many times loop gain may be increased before reaching a stability limit
• Phase margin is the phase shift that is added to the loop phase shift at the gain crossover frequency to make a total phase shift of 180 degrees.

The robustness map presents a range of reasonable tuning parameters from which the usercan select. Users can simply click a point to see a set of tuning parameters and view the responses. Significant variation in the setpoint overshoot in shown below for different tuning points on the robustness map.

The Solution for the On-demand workshop  in Chapter 4 may be accessed at the ControlLoopFoundation web site. The solution is a YouTube video that include a demonstation that show how the Robustness map is used in tuning a control loop.

Continue reading »]]> When the control objective is to maintain the controlled parameter of an integrating process at setpoint under all operating conditions, then using PI (proportional-integral or proportional plus reset) controller, it is possible to automatically compensate for changes in disturbance inputs and maintain the controlled parameter at setpoint. However, the application of PI control to an integrating process can be quite challenging. Boiler drum level control, illustrated below, is an example of an integrating processes where the control objective is to maintain the level at setpoint.

At key factor in commissioning PI control of an integrating process is to correctly determine the reset that is applied. Once the reset is established then the proportional gain can be adjusted to achieve the desired response.

When working with an integrating process, IMC and Lambda tuning rules agree on the setting of the reset:

As discussed in Chapter 4 of Advanced Control Foundation – Tools, Techniques, and Applications, Model based tuning using the IMC or Lambda rules has some advantages over other tuning rules such as the SIMC method for the tuning of an integrating process.

If you would like to gain experience tuning PI control of an integrating process, then I would encourage you to explore the three element drum level control workshop that is contain on the web site for Control Loop Foundation – Batch and Continuous process. A YouTube video showing the workshop solution can be accessed by selection the Solution tab.

Continue reading »]]> Intermediate liquid storage tanks (surge tanks) are commonly installed between processing areas of the plant. Under normal operating conditions, these storage buffers allow each process area to be operated independently of the other process areas. Any imbalance in the area production rates within a plant will be reflected by a change in level of the surge tanks between process areas. When a downstream process area is made up of a continuous process, then its throughput may be automatically adjusted based on the surge tank level, as illustrated below.

To avoid abrupt changes in production rate in the downstream process, the level controller (LC 202) may be configured for proportional-only control, with the bias setting based on the normal plant production rate. As described in Chapter 11 of Control Loop Foundation – Batch and Continuous Processes, the range over which this floating-level control adjusts the downstream flow is determined by the controller gain and bias.

For this tank level example where the PID output directly adjusts the tank outlet flow valve, the BIAS value would be set equal to the valve position needed to achieve normal flow rate through the tank. The proportional gain in this example may be used to determine the outlet flow rate change necessary to compensate for a change in level. For example, if the proportional gain is set to 1 and the BIAS is set to 50% (assuming an outlet valve position of 50% is normally needed to maintain the tank level), then the control output to the valve will be full open if the level reaches the upper limit and full close if the level reaches the lower limit of the transmitter range. If the proportional gain is set to 2 with a BIAS of 50% and a setpoint of 50%, then the control output would be full open when the level reaches 75% and full closed when the level reaches 25%.

The proportional gain in this example determines over what range the level is allowed to vary (float). The use of proportional-only control for tank level control is often called floating level control. That is, the level is allowed to vary or float within a certain range, and the amount it floats is determined by the proportional gain. In this case, the objective is to take full advantage of the surge capacity of the tank, that is, to not immediately pass upstream process throughput changes to the downstream process. If applied in this manner then proportional-only control can be quite effective.

Continue reading »]]> Older control systems may not include tools that allow the PID tuning to be automatically established. In such cases, the following procedure outlined in Chapter 12 of Control Loop Foundation – Batch and Continuous Processes may be quickly applied to determine the tuning of a self-regulating process for PI control.

In Chapter 4 of Advanced Control Foundation – Tools, Techniques and Applications we address on-demand tuning. More detail is provided in the Technical Basis section on tuning rules that are commonly utilized in the process industry. One of these latest techniques is the SIMC method for PID controller tuning. This technique was developed by Professor Sigurd Skogestad, Norwegian University of Science and Technology, using direct synthesis and then applying heuristic evaluation of how to modify the formulas to improve response on disturbance rejection. One of his objectives was to develop simple and easy-to-memorize rules. The resulting formulas for a self-regulating process and PI control are shown below:

The rule integrating process and PI controller are equally simple and easy to apply:

If you want to learn more about Skogestad’s work on controller tuning then I would encourage you to read a paper he presented on SMIC tuning in one of the plenary sessions at the IFAC PID’12 conference. I had a chance to sit in on his presentation and to talk with him at the conference. The rules he developed can be easily applied with good results on a broad range of applications.

Continue reading »]]> Control system manufacturers have adopted different technologies to implement on-demand tuning. Because these tools are designed to work in a noisy process measurement environment, the tuning results are usually better than those achieved through manual tuning techniques. To illustrate the features that may be included in on-demand tuning capability, the on-demand tuning capability of the DeltaV control system is examined in Chapter 4 of Advanced Control Foundation – Tools, Techniques, and Applications. Other tools may differ in the technique used to introduce a change in the manipulated process input and the methods used to identify the process gain and process dynamics and to provide the recommended tuning.

The auto-tuning user interface is used by a variety of people in the plant and must be designed to be easy to use. The interface will typically include a trend of the controlled parameter PV, the setpoint SP and output of the PID block. Also, other information may be automatically displayed that is needed in loop commissioning, such as the current tuning of the PID. In this example, the PID tuning process is initiated using the Test button. Once the tuning begins, the output of the PID is changed automatically from its initial value by the default step size as illustrated below.

Based on the process gain and dynamics identified through the automated test, the auto-tuning application calculates a recommended setting for the PID proportional, integral, and derivative gains. To use the tuning determined by the application, the user simply selects the Update button, and the tuning values are automatically written to the PID block. Thus, with two clicks of the mouse, a control loop can be tuned.

The workshop for Chapter 4 illustrates some of the features that may be found in an on-demand tuning product. An example heater process is used in this workshop. A YouTube video that shows the workshop solution may be viewed by accessing the book’s web site.

Continue reading »]]> After the process characteristics are identified, the information is then used to calculate the recommended loop tuning. The tuning rules used to derive controller parameters from process characteristics can be chosen as appropriate for the specific process and the desired speed of response. As addressed in Chapter 4 of Advanced Control Foundtion – Tools, Techniques, and Applications, the response to a setpoint or load disturbance change is determined by the controller tuning and can be classified as follows:

• Overdamped response – Controlled parameter is gradually restored without overshooting the setpoint value. In most cases an overdamped response is the best response for varying operating conditions that impact process gain and response time.
• Critically damped response – Controlled parameter is restored in minimum time without overshooting the setpoint value. While this response minimizes the time to respond to a change in setpoint or process disturbance, unstable control can be observed with changing operating conditions that impact process gain or response time.
• Underdamped response – Controlled parameter overshoots the setpoint value but eventually settles at the setpoint value. This response may minimize time to get back to setpoint but at the expense of stability and the controlled parameter overshooting to setpoint when responding to a change in setpoint or change in load disturbances.

Once the reset gain of the PID is set based on the identified process deadtime and time constant, it is possible to go from an overdamped response to a critically damped response or to an underdamped response by adjusting the PID proportional gain. Some on-demand tuning applications have a selection that changes the proportional gain to achieve a specified speed of response, for example, a fast, normal, or slow response.

The best tuning for a particular loop depends upon the control objectives. If the setpoint is constantly being changed, the loop’s response to setpoint changes would be a key factor in evaluating its performance. However, if the loop setpoint is always maintained at a fixed value, the response to disturbance inputs would be more important. If the response to both setpoint and disturbance inputs is important, tuning must be set to achieve a balance in control response. In many processes it is desirable to have an overdamped response. To achieve that response, the proportional gain used in the PID is typically much less than that required for an underdamped or critically damped response.

Continue reading »]]> When the on-demand tuning application resides in a DCS workstation, communications between the controller and the workstation can introduce variable delay and jitter in the observed process response. As addressed in Chapter 4 of Advanced Control Foundation – Tools, Techniques, and Applications, the process may be more accurately identified by distributing the on-demand tuning functionality between the workstation and the controller as illustrated below.

Capturing process dynamics in the field (controller or device) instead of using test data transferred to the workstation eliminates any variation introduced by communications and thus allows better process identification, particularly for the fastest loops. For example, this is the way on-demand and adaptive tuning capability are structured in the DeltaV control system. When the process response is captured in this manner, the tuning rules calculation may reside in the workstation. Capturing the process response in a controller or field device requires only simple mathematical operations and is no more CPU resource-consuming than the control algorithm. Thus, switching over to the identification algorithm does not result in an overall load increase in the controller or field device.

Most on-demand tuning applications are designed to provide feedback indicating when process testing is in progress. Also, safety features are common and provide quick response to process conditions, such as too large a deviation of the controlled parameter from setpoint, which can indicate abnormal operating conditions, limit conditions, or significant process disturbance. These conditions will cause process identification to abort and normal control to resume.

Continue reading »]]> During the commissioning of feedback control based on the PID algorithm, the PID tuning parameters (that is, the proportional, integral, and derivative gain) must be set to specific values to achieve the best controller response to setpoint and disturbance input changes. To minimize process variations and the response time to setpoint and disturbance input changes while providing stable operation, the PID tuning should be based on the observed process gain and process dynamics for each control loop. The instrumentation engineer involved in plant commissioning may not have correctly set the loop tuning, or the process operating conditions may have changed, and as a result plant operation does not achieve maximum efficiency. Fortunately, most distributed control systems (DCS) sold today include on-demand tuning support that can be used to automatically establish the correct loop tuning.

In most cases the on-demand tuning capability of a DCS, also known as auto-tuning capability, is based on the identification of the process model that matches the observed process response for a step change in a manipulated input. To identify the process response, it is necessary to test the process by initiating a change in the manipulated process input. As addressed in Chapter 4 of Advanced Control Foundation – Tools, Techniques, and Applications, the manner in which this is done varies with the on-demand identification technique used by the DCS manufacturer. The size of the change introduced in the manipulated process input must be large enough to easily distinguish the process response from any process noise that may be present. For a self-regulating process, the process response is most often approximated as first order-plus-deadtime, that is, the response is characterized by gain, time constant, and deadtime as illustrated below.

The response of an integrating process can be characterized by the process integrating gain and deadtime as illustrated below.

Based on the identified response, tuning rules are applied to determine the recommended PID tuning.

Continue reading »]]> Many control system manufacturers and third party companies now offer tools that are designed to monitor process control and instrumentation performance. In addition to providing on-line information that summarizes control utilization, the user may access other information that may be helpful in determining the cause of, and solution to, poor control utilization. In section 3.4 of Advanced Control Foundation – Tools, Techniques, and Applications the performance monitoring capability embedded in the DeltaV control system is used to illustrate some of the features that may be found in commercial performance monitoring tool. For example, when the performance monitoring application is first opened an overview of the plant may be displayed as shown below.

From this overview, the user may select to see a performance summary for an individual process area, unit, or cell in the plant.

The workshop included in Chapter 3 is designed to illustrate features that are typically available in commercial performance monitoring products. The process areas and exercise for this workshop may be viewed by accessing the book’s web site. A YouTube video that shows the workshop solution may be viewed by selecting the Solution tab.

Continue reading »]]> In some cases, the manipulated parameter of one control loop can impact the controlled parameter of another control loop as illustrated below.

To allow both loops to operate in automatic mode, loop interaction is most often addressed by simply detuning one of the control loops by reducing the proportional gain. The valve (or another final control element) associated with the detuned loop will change position slowly. Thus, the two loops will tend not to interact but at the expense of higher variability in the de-tuned control loop.

When model predictive control, MPC, is used to address in a control application, the interaction of the manipulated inputs and controlled outputs may be addressed automatically while providing the best performance for all control loops. As addressed in Chapters 11 and 12 of Advanced Control Foundation – Tools, Techniques, and Applications, the impact of each manipulated input parameter on each controlled output parameter is identified by the step response model used in MPC block generation. Thus, any interactions are automatically compensated for when using model predictive control.

Continue reading »]]> One of the most common ways of addressing multiple process inputs is split range control. A splitter block, which appears as one valve to the PID block, can be used in the control strategy to define a fixed relationship between the controller output and each manipulated process input. In Advanced Control Foundation – Tools, Techniques, and Applications, Chapter 3 we used the following example in the discussion of split range control. To achieve consistent control behavior, the splitter block setup must account for the gain associated with each process input. Unfortunately, in many installations the split range setup was arbitrarily defined during commissioning as an equal split among the process inputs. Such a setting can result in sluggish or unstable operation when the controller output transitions from one input to another input.

The manner in which the splitter block setup accounts for the gain associated with each process input may be illustrated by considering a steam header pressure control application. One or more power boilers may be used to meet the process steam requirements of a plant. Also, this steam may be used in turbine generators to meet some or all of the plant requirements for steam. Steam turbines used to generate electricity are powered by high pressure steam. For example, in newer installations, the supply header may be at 1475 psi. The pressure of the steam is reduced as it flows through the turbine, allowing lower pressure steam to be extracted at various points. This lower pressure steam may be used to meet the plant’s process steam demands. For example, at one point, the extraction may be regulated to meet the steam demands of a 400 psi header. Extraction valves associated with the turbine are automatically regulated to maintain the lower header pressure constant.

To allow the plant to continue operation if a turbine or generator fails and must be shut down, pressure reducing valves (PRVs) between the high pressure header and the lower pressure header can be adjusted to meet the lower pressure header steam demand and to maintain the lower header pressure constant. This can be accomplished by using a splitter block in conjunction with a PID block to adjust the pressure-reducing valves as illustrated below.

If the valve sizes or operating conditions differ from loop to loop it is necessary to characterize the splitter to compensate for these differences through the setup of the characterizer block. When a loop that includes a split range output has low utilization, examine the assumptions made in the split range setup and, if necessary, modify the setup to provide a constant process gain as seen by the controller.

Continue reading »]]> From a control perspective, it is highly desirable that the process gain is constant. If the process gain is constant, then the same proportional gain can be used over the entire operating range of the control loop. Using a control valve as an example, if the valve characteristic was not properly selected based on the process requirements, the valve installed characteristic could be non-linear.  In Chapter 3 of Advanced Control Foundation – Tools, Techniques, and Applications the following example is provided that shown the impact of a non-linear valve installed characteristic. In this example the process gain varies from 0.5 to 4; that is, the process gain changes by a factor of eight.

For such non-linear installed characteristics, the impact of the changing process gain with valve position can often be seen by trending the controlled parameter (PV), setpoint (SP) and controller output (OUT) to the value. If valve is found to have non-linear installed characteristic then stable operation can be achieved by tuning the control loop in the high gain region, but doing this will result in slow response when the process is operating in the low gain region. To compensate for the changes in process gain, a characterizer block (SGCR) can be installed between the PID and Analog Output blocks as illustrated below.

Because of the time and understanding that are necessary to install, commission, and maintain a control loop that includes a characterizer, this approach is not commonly found in the process industry. To improve control performance, the valve installation may be modified to provide linear installed characteristics. For example, the valve trim can be modified or a digital valve positioner installed that supports characterization.

Continue reading »]]> When a control loop is placed in automatic control, it is easy to detect if a valve or actuator is not responding correctly to the control system. As discussed in Chapter 3 of Advanced Control Foundation – Tools, Techniques and Application, this can be done by observing the response of the controlled parameter to control system changes in the PID output as illustrated below.

The most common problems in commissioning a control system can often be traced to a valve without a positioner, or to a positioner thathas not been properly installed or has malfunctioned. The rule of thumb is that to achieve best control performance, all regulating valves should be equipped with a properly functioning positioner. Without a positioner, the level of control performance that can be achieved is very limited when a valve is sticking—which is inherent in most valves.

The cyclic behavior caused by a sticky valve (with no positioner) or a malfunctioning positioner cannot be eliminated through tuning. Changes in tuning will only impact the period of the cycle. The only way to eliminate this type of behavior is to install a valve positioner and ensure that it is functioning properly

Continue reading »]]> A Loop tuning has a significant impact on control loop utilization and performance. Controller tuning parameters can be easily changed over a wide range to adjust loop operation for various process requirements. However, some expertise in tuning is required even when advanced control tools are used, and there is a chance that the behavior of a particular process will be unaccounted for in the tuning that was selected for the control loop. These are important reasons for considering loop tuning as a primary suspect in degraded performance.

During the design of a new process area or plant, initial loop tuning can be specified based on the type of measurement associated with the control loop, that is, whether pressure, level, temperature, flow, or some other process variable to be controlled. In most cases, these default settings may be close enough to support the initial startup of a new process area or plant. However, to minimize process variations from setpoint and to minimize the response time to setpoint and disturbance input changes while providing stable operation over a variety of operating conditions during or after startup, it is necessary to set the PID tuning based on the observed process gain and process dynamics for each control loop. Unfortunately, the instrument engineer involved in plant commissioning or plant operations may not have the opportunity to tune loops and, as a result, plant operation does not achieve maximum efficiency.

In an operating plant, various approaches can be taken by plant personnel to modify control loop tuning to improve control performance. The most direct and reliable way to manually or automatically establish PID block tuning is to calculate the tuning based on the identified process gain and dynamics as described in the following sections. With older control systems, the analog single loop controllers and programmable logic controllers (PLCs) used may not include tools that allow the PID tuning to be automatically established based on the observed process response to input changes. As is address in Chapter 4 and 5 of Advanced Control Foundation – Tools, Techniques, and Applications, most modern control system provide some to automatically establish loop tuning, as illustrated below.

When selecting the PID gain, it is important to consider that the process gain can change with the operating conditions. To ensure stable operation over the entire operating range of a control loop, the normal practice is to establish the PID tuning when the process is operating in the region with the highest process gain. As a result, higher variation in the controlled parameter can be seen when the process transitions to the low gain region of operation.

Continue reading »]]> One of the most obvious reasons for poor control utilization is a broken or unreliable transmitter or sensor. As addressed in Chapter 3 of Advanced Control Foundation – Tools, Techniques, and Applications, if the control system is designed to be consistent with the international function block standard, IEC 61804, the status attribute is always communicated with the measurement value as illustrated below and provides a direct indication of measurement condition. The percentage of time the measurement had a status of Bad, Uncertain, or Limited can be used to determine measurement health. Thus, the status of a function block is the basis for evaluating measurement health.

The status attribute that accompanies a measurement value provides an indication of the quality of the measurement. For a measurement input provided by a traditional transmitter, the status attribute is determined by the analog input card. For a digital value provided by the transmitter, the status attribute can be provided by the transmitter. The status attribute value consists of eight bits. The most significant two bits are used to indicate the quality of the measurement classified as: Bad, Uncertain, Good – Control, and Good – Measurement. The last two bits of the status attribute, the lower-priority bits, are used to indicate high or low limit, no limit, or constant. The four middle bits of the status attribute are used to show why the quality is good, bad, or uncertain. The health of a measurement may be assessed based on the percent of time its status was Bad or Uncertain or a Good but an active limit condition.

After a measurement value and status are generated by the analog input block, the other blocks that use this measurement input process the status and use it in some way, and propagate a status to the downstream blocks. If the input block status is Bad, the Actual mode of the block that uses a measurement provided by the input block is unable to achieve an Automatic mode of operation and thus the Normal and Actual mode attributes will indicate that control is not being utilized as designed.

Most problems associated with a measurement can be quickly addressed by calibration or replacement of the transmitter.

Continue reading »]]> When the utilization of a control loop is found to be low during normal plant operation, the measurement or control problems that prevent the control loop from functioning in its normal (designed) mode of operation should be further explored. In Chapter 3 of Advanced Control Foundation – Tools, Techniques, and Applications we outline the steps involved in resolving problems that impact control utilization as illustrated below. To achieve full utilization, effective communication between maintenance and operations is important.

When a control loop is routinely not operated in normal mode (which is most often automatic mode), further investigation is needed to determine the cause of the poor control utilization. Some of the most common problems that can impact control loop operation in automatic mode are:

• Measurement health, i.e., quality of the measurement
• Incorrect tuning
• Valve/actuator malfunction
• Changing process gain
• Split range setup
• Process gain
• Process dynamics
• Loop interaction

Control performance tools provide information that guides the user in solving the performance problem. However, where there is no clear indication of the source of the problem, it may be necessary to test and observe the loop operation in both manual and automatic modes.

Continue reading »]]> It can be shocking to examine control utilization in a plant where performance tools have not been installed and/or not used. For example, in the mid-1980s a control survey was conducted at a major pulp and paper plant where the control system had recently been updated to the latest distributed control system. At that time performance tools were not available to automatically evaluate control system performance. However, it was possible to manually evaluate control utilization using a snapshot of the plant operation. The results of this survey are documented in Chapter 3 of Advanced Control Foundation – Tools, Techniques, and Applications as shown below.

After management saw the results of this survey, an instrumentation team was formed to investigate loops that were not running in their normal (design) mode. This team was responsible for making sure measurement, control valve, and process problems were addressed in a timely fashion. The resulting reduction in variability led to significant improvements in plant throughput and product quality. Two years later the plant set a new production record.

Even today, similar problems can be found in process areas equipped with the latest control systems and field instrumentation. For example, a survey of seven areas in a petrochemical complex was recently conducted using performance monitoring tools that were embedded in the control system. A summary report of the control utilization for this plant is shown in below.

Once the plant management at this large petrochemical complex became aware of the low control utilization in key processing areas, manpower and funding were provided to investigate and correct the measurement, control, and process problems that were preventing the operator from using the control as designed. This led to significant operational improvements.  These examples demonstrate that the path to improving control performance should start with an assessment of control utilization. The automatic collection of control utilization statistics by the control system is of major benefit in identifying problems in measurement, control, or process design and operation.

In plants that do not have a control system that provides embedded control performance monitoring, this capability can be layered onto the existing control system. When a control loop is identified as not being fully utilized, the reason for poor utilization should be found and resolved. Once the issues of control utilization have been addressed it is possible to start examining the variability in control parameters, expressed by its standard deviation. Where the variation is large enough to impact plant production or product quality, the source of the variation should be investigated and steps taken to minimize it.

Continue reading »]]> A plant operator is usually responsible for managing one or more process areas, with (potentially) hundreds of control loops and process measurements in each area. The operator’s job is made easier if the control loops in these process areas are in automatic mode and can compensate automatically for disturbances and maintain the operating conditions the operator specifies through the control loop setpoints. Thus, the operator will place a control loop in a manual mode of operation only if there is a measurement or control problem that prevents the control loop from satisfying operational or performance requirements. Therefore, in evaluating measurement and control performance, a good starting point is to quantify current and average control loop utilization in each process area of the plant.

As addressed in Chapter 3 of Advanced Control Foundation – Tools, Techniques and Applications, most performance monitoring tools are designed to provide both an instantaneous value and an historic view that summarizes control loop utilization by process area. If the utilization is lower than some established limit, the user can select a detail view of the process area that shows the utilization for each control loop. Control utilization can be determined by comparing the Actual and Normal mode attributes of the mode parameter. The mode parameter of the PID block determines the source of the block setpoint and the source of its output. This parameter is a critical part of the operator’s interface to the PID block and consists of four mode attributes:

• Target – The mode of operation requested by the operator
•  Actual – The mode of operation that can be achieved based on the target mode and the status of the PID block inputs
• Permitted – The mode(s) available to the operator for a given application
• Normal – The normal operating mode of the PID

The Actual mode attribute can take on a value other than that specified by the Target mode attribute. This occurs when an internal condition or input status indicates that the mode of operation requested by the operator through the Target mode attribute cannot be achieved. The workshop on evaluating control system performance in Chapter 3 of Advanced Control Foundation – Tools, Techniques and Applications illustrate this behavior. The solution for this workshop may be viewed on the book’s web site.  Under normal operating conditions, the Target and Actual mode attributes match. When the Actual mode attribute changes to Local Override (LO), the output tracking or an internal application such as autotuning is setting the block output. Similarly, if the path to the process is lost, the Actual mode attribute changes to Initialization Manual (IMan). When the Actual mode attribute of a PID is IMan, the path to the process is incomplete. The path to the process can be incomplete if a downstream block is taken out of Cascade mode. The path to the process can also be incomplete because of a physical condition such as the hardware failure of a downstream block or an actuator failure that is manipulated by the PID.