Setting the contours in the industry

Assignment of ExperTune


(most of the pictures, 114 images, see below)

The program ExperTune exists in the world as a package of additional, built-in opportunities for tuning technological control loops. It is designed for automated analysis, diagnostics, adjustment and optimization of PID control circuits for industrial facilities. The program is integrated with the APACS + / QUADLOG / ProcessSuite system (the brand was absorbed by Siemens in 2007, is no longer supported), also Expertune package supports most modern SCADA platforms.

According to the available estimates, 75-80% of the control loops in industrial enterprises are more variable in automatic mode than in manual control mode. At the same time, one third of them, under load or setpoint (setpoint) changes, produces cyclic oscillations due to different non-linearities: hysteresis, delay, valve sticking, nonlinearity of the process. Another about a third generates cyclical fluctuations due to poor tuning of the regulators, while the others cause this design error.

Modeling of control loops

One of the main and effective tools of ExperTune is the mathematical and software modeling of the technological process and the control loop.

ExperTune provides the ability to dynamically simulate and simulate the operation of a closed control system based on the received data on the behavior of the control object after a single test impact on the process. This allows the user to check the efficiency of the recommended settings on the closed-loop model before they are loaded into the PID controller, which is especially useful if the recommended ExperTune options do not completely satisfy the user. In this case, he can change the settings offered by the program and check them on the model without interfering with the real circuit. To construct the model, effective methods are used for frequency analysis of the results of the loop response to a step or impulse disturbance at the input.

ExperTune automatically selects the most suitable model of the technological process-the load of your circuit: a 1st or 2nd order process with a delay (deadband), a 2nd order process with overshoot and delay, integrating the process with delay.

Automated setting of PID controllers

ExperTune simplifies and automates the process of data collection, analysis, diagnostics and optimal adjustment of PID control loops directly on the control object. To do this, the user only needs to specify the control loop and then follow the instructions of the program. ExperTune asks to give the circuit a single "push" or pulse, and measures the response of the object to an external action (changing the setpoint SP in a closed or control signal CO in an open loop). After the end of the transient process, the program uses the measured data to build a mathematical model of the process and calculates the PID control parameters that provide the user with the characteristics of the loop. At the same time, the program evaluates the expected improvement in the quality of the loop function with the new settings. If the user approves the program recommendations, the new parameters can be loaded into the regulator and returned to the previous automatic control mode. The time saving of each loop setting is from 2 to 6 hours.

With a DDE / OPC server, you can monitor, configure, and analyze multiple circuits at the same time.

Stages of diagnostics and optimization of the control loop

The complete optimization of the control loop with the help of ExperTune includes the following 5 steps:

1. Statistical analysis and diagnostics on the basis of real data of the normal operation of a closed loop in the system.

2. Frequency analysis (spectral density of signals) to detect hidden cycles in the circuit.

3. Analysis of the operation of the control valve to detect and correct such defects as hysteresis and sticking.

4. Check for non-linearity and linearization of the control loop.

5. Determine the optimum controller settings.

Statistical analysis and diagnosis of "sticking" of the valve

The statistical analysis of the circuit operation with the changed settings on the model allows one to assess how much the process characteristics such as standard deviation, variability and load on the valve (the intensity of its movements) are improved. The results of analysis of the real circuit and its model are presented in numerical and graphical form - as a histogram of the distribution of the values ​​of the regulated variable PV. This histogram, in particular, helps to detect such defects in the operating circuit as "sticking" of the valve, which is often the cause of poor regulation and leads to cyclic fluctuations in the circuit. ExperTune calculates a numerical estimate of the degree of "sticking" of the valve existing in the analyzed loop. If the measured sticking ratio of the valve is more than 1%, then its repair can give a big effect - increase the speed of the circuit and eliminate the cyclical fluctuations

Analysis of the stability of the contour

In addition to the controller settings, ExperTune calculates and displays a graphical representation of the "stability region" of this control loop, characterizing the degree of stability of the closed loop with the current settings. The current state of the process stability in the circuit is represented by a point on the plane in the coordinates X = gain factor PG, Y = dead time in sec. If the working point lies deep enough within the stability region, then we can say that this contour is very likely to be "stable", that is, relatively small changes in the characteristics of the control object can not make the circuit unstable and cause self-oscillation.

Dimensions of the "stability area" depend on the current settings of the PID = regulator. Thus, the stability graph gives the user a graphical representation of the degree of "instability risk" of the circuit at the given controller settings.

Frequency analysis

Frequency (spectral) analysis, that is, analysis of the frequency spectrum of energy distribution of signals, is a powerful tool that allows you to detect hidden cycles in the operation of control loops, trace and find a loop that causes cyclic fluctuations in the operation of the entire installation, and cross-correlation calculations help to detect interdependent pairs of contours.

Frequency analysis also allows you to answer the following questions:

• Has the control loop performance improved with the new settings?

• Has the service life of the valve increased?

• Are there any hidden cycles in the noise of the circuit?

There is no need to test the circuit by the "jump" method of the setpoint or load.

To reveal hidden cycles, graphical diagrams of the spectral density of the PV signals and the output of the CO regulator are used. Cross-correlation graphs of CO-PV allow to determine the presence of mutual influence between two circuits.

In order to separate the interconnected flow and pressure control loops, the flow loop must be adjusted so that it works more slowly than the pressure loop, but at the same time ensures effective control. ExperTune automates this process based on frequency analysis of the response of both circuits to a "jump" in the load. In this case, special parameters are calculated - the "relative" reaction time (RRT) of the first and second control loops, which in the cascade should differ by at least three times.

Longer service life of control valves

Based on the current parameters of the controller settings, ExperTune determines and predicts the frequency of movement and deterioration of the valve, and then recommends filters and new settings that will increase the life of the valve.

For this, the program calculates the coefficients characterizing the intensity of the valve movements and the frequency of its reversal (changes in the direction of motion) for the current and new (optimal) regulator settings, compares them and determines the optimal filter.

With the help of the ExperTune analysis tools, the user can change the PID controller settings and in a few minutes see how this will affect the life of the valve. Applying the filter to an adjustable variable (PV) or changing the integration time can increase the valve life several times.

ExperTune optimizes the PID controller settings and the adjustable PV variable filter parameters so as to extend the life of the valve by reducing the intensity of its movements and the number of reverses that are performed, but at the same time maintaining the quality of the control and the stability of the circuit.

A properly selected filter increases the effective operation time of the valve between repairs, reduces operating costs and reduces the level of generated noise.

After determining the filter type and regulator settings that are optimal for the valve operation, the program simulates the operation of the circuit to ensure that the quality of the control and the stability of the circuit are not affected.

Calculation of the SP job filter

Normally, the regulator is set up so that sufficiently rapid changes in the output signal of the regulator reduce possible fluctuations in the load. However, in this case, often the effect of overshoot (ejection) of the PV with a step change in the setpoint SP. To eliminate this defect, ExperTune calculates the parameters of the lead-lag filter, which is placed at the regulator setpoint input. If there is no such block in the system, ExperTune gives the user the equations for programming the filter.

With this filter, the loop can respond quickly to load changes with good control quality. The setpoint filter provides a high-quality operation of the regulator during the start-up of the process unit and changes in its operating modes, which is usually accompanied by a change in the setting values, while maintaining optimal settings for a normal, stable control mode.

Linearization of nonlinear circuits

In the case of a non-linear process, the control loop can behave very differently at different sections of the load scale (adjustable variable), for example, stably operate in one part of the range and generate cyclic oscillations on the other. To eliminate non-linearity in the control loop, a special software Linearization Block is usually added to the controller.

The program ExperTune calculates and gives the user all the data necessary for programming the linearizer unit in the controller so that the circuit with the found settings works equally effectively on the entire load range. For linearization, a linear-piece approximation or a hyperbolic function is used. These possibilities can be used, for example, for the linearization of cost control loops, secondary (slave) contours of cascade controllers, and also in any nonlinear circuit with a variable setpoint.

Universal unit for linearizing the control loops of the pH-value of hydrogen

The circuits of the pH regulators are non-linear in nature, and there are cyclical oscillations around the setpoint value. The program ExperTune linearizes any pH-loop, resulting in regulation with little or no oscillation. This improves the quality of the product, reduces the consumption of reagents and helps to keep the pH within specified limits.

Processes with inverse response to the control action

There are especially difficult for the regulation of technological processes, where when the output of the regulator changes the controlled variable begins to change in the direction opposite to its required final value. ExperTune helps to easily simulate, configure and optimize such processes. In this case, the program automatically determines the necessary values ​​of the settings parameters - negative lead and integration time constant for this inverse process.

Optimization of the outline and the Summary table of settings

To optimize the control loop, it is often necessary to find a balance - a compromise between the performance of the contour, its stability and the intensity of the movements of the valve. Changing the settings of the PID controller changes the speed of the loop response to fluctuations in load or setpoint. More aggressive settings improve performance, but can also increase the sensitivity of the loop to changes in the characteristics of the control object, that is, to reduce its stability. In the program ExperTune, the user can enter the desired coefficient of "stability margin", which will be used to determine the optimal settings.

ExperTune forms and stores a summary table of all the options for the PID controller settings obtained during its testing and simulation. This table helps to find the best combination of settings for those "volatile" circuits that behave differently when the process conditions change. The program analyzes the data of several tests and finds the most suitable, conservative setting of the regulator, which may be the only one necessary for stable contour operation. Therefore, it is sufficient for the user to perform several contour tests, and then the program will automatically generate a table, select the most favorable and average values ​​of settings and parameters of the contour model. An example is some unbalanced temperature control loops, which are triggered by heating faster than cooling. Having ExperTune, it is enough for the user to perform two test "push" of the circuit in the manual mode of the controller, first in the heating direction, and then in the cooling direction. The program will print out and save the settings for both cases, and the user will be able to choose the preferred option.

Analysis of time relationships of the circuit

ExperTune analyzes the ratio of various time parameters that determine the dynamic properties of the control loop. For example, the cycle of polling and data processing in the controller should always be substantially less than the dead band of the loop. If this is not the case, then reducing the cycle time can significantly improve the loop performance. The following parameters are analyzed and evaluated:

• Data processing cycle

• Filter time constant

• Time constant of differentiation

• Dead time zone

• Equivalent deadband (for the 2nd order process)

• Integration time

• Relative contour response time (RRT)

Compensatory feedforward regulation

If an external factor that causes changes in the adjustable PV variable is known, for example, another technological variable FW, which thus also serves as an "implicit" load for the loop, it is possible to calculate the control action in the control loop that will compensate for the negative effect of the FW changes on the variable PV. To do this, it is necessary to measure this "influencing" load variable FW and have a compensation calculation unit. The program ExperTune, firstly, finds such an "influential" variable FW and, secondly, creates for the user on its workstation the algorithm and the program of work of the Compensator, which is loaded into the controller and is a program for calculating the compensating control action. The user does not need to do any additional tests or checks, and you can immediately use the received Compensator at the facility.

This method of regulation can significantly reduce the impact of random load fluctuations, while simultaneously increasing product quality, equipment life and reducing variability (variability) of the contour.

Unstable processes

ExperTune works efficiently if the model constructed accurately reflects the behavior of the real control loop. The accuracy of the model depends on the reliability of the data obtained when testing the object by the "jump" method at the input. To ensure the reliability of the data, testing should be carried out at a "quiet" state of the system (that is, in the absence of significant load fluctuations) and in the linear range of its operation. However, these conditions can be difficult to observe in the case of unstable processes subject to random load fluctuations or processes that react differently to positive and negative changes in the control action. In such a situation, it is advisable to use more sophisticated methods of modeling and adjusting with the involvement of the company's engineering experts.

 The program provides the ability to collect and record on the disk all the necessary information that can be sent to the company for analysis and solving complex problems.

Developed reference and information system and training system (in English)

In addition to the usual access to information through the Help menu, it is enough to place the cursor on the interesting element of the image on the screen and press the F1 key to call the help.

To call the training system, click the "tutorial" link in the help text or select "tutorial" from the window menu.


ExperTune generates a complete report on all operations and analysis results, tuning, optimization and diagnostics of each control loop, including all graphic and numerical data, loop performance and filter parameters of the setpoint. The report is issued as a Microsoft Word document.

Connecting the program to the controllers

The program runs on the Windows NT / 2000 platform and interfaces with all common types of controllers through the available drivers.

The ProcessSuite software package of the APACS + / QUADLOG system fully supports the work with the ExperTune program. For this purpose, in the environment of the operator-technologist Vision Framework R5 on the operator's stations of the system, there is an ExperTune call from the standard "PS-screen" window of any control loop configured in the system.

Sources of how to use ExperTune

• The operation of the control loops is optimized

• When configuring each circuit, the time savings range from 2 to 6 hours

• Improves product quality

• Losses and quantity of marriage are reduced

• Electricity is saved

• User-friendly and simple user interface of the program with a developed reference and training system.


Variability of the contour - Generalized statistical characteristic of the contour, calculated in the statistical analysis of test data (through the variance of the test sample). It characterizes the average value of the variance of the values ​​of the controlled variable with a stable state of the contour. We can say that variability is a measure of "non-ideal" regulation.

Loop Performance - the effectiveness of the application of the control loop in the process, affecting the quality and cost of production. It is determined by the speed and quality of regulation in the circuit. Ultimately, this indicator is proportional to the economic efficiency of the circuit.

The trend data window ("Time data for ...", Time Plot) is a window for displaying, editing and analyzing contour testing data. The trends in the variables PV and CO in this window always represent those initial data that the program uses for all subsequent calculations and analysis, including the definitions of the mathematical model of the process and the settings of the controller. These trends, i.e. test data, can be edited.

Relative Response Time RRT (Relative Response Time) - the relative rate of response, the response rate of the control loop to the disturbance. The smaller the RRT, the faster the performance and vice versa, the larger the RRT, the slower the circuit will work. This indicator is useful for different tasks of comparing contours, models and settings.


Optimization of PID control loops allows to significantly improve the efficiency of the control system. When the circuit operates optimally, stability of its operation increases and accordingly the variability decreases. This ensures operation in a mode that meets the requirements of the technical regulations.

Optimizing the circuit is not only setting the parameters of the PID controller.

A typical control loop includes: a variable-variable process, a PID controller, typically an air-electric converter and a control valve. For optimal control of the technological process, it is necessary that all these components work well. Therefore, before configuring the path, you should check the operation of each of its elements and make sure that they all work correctly. The regulator setting is the last step in the complete optimization of the loop.

With the optimum settings, your process equipment will be used with maximum efficiency.

To evaluate the result of the performed loop optimization, the program ExperTune calculates two generalized indicators:

• Factor of increase in "productivity", efficiency of the "Performance" loop (speed and quality of regulation), which shows how much better the PID controller will work out load disturbances with new settings. This ratio is directly proportional to the expected savings in cash.

• Coefficient (degree) of stability "Robustness" of the closed loop to changes, drift of dynamic parameters of the technological process (load), characterizing the stock of its stability and protection from the appearance of self-oscillations. Statistical analysis of ExperTune allows to detect and estimate the variability of the contour


1. Statistical analysis and diagnostics according to the data of normal stable contour operation

2. Frequency analysis (spectral density of signal power) for detecting hidden cycles

3. Diagnosis and elimination of control valve problems: adhesion, hysteresis, size mismatch (sizing)

4. Checking for nonlinearities and linearizing the contour

5. Determination of the optimal PID settings and filter

Next, a step-by-step procedure for diagnosing and optimizing your control loop is described in more detail.

Conduct a survey of operational and engineering personnel on existing problems and comments on the work of the circuit. If it is found, in particular, that the circuit is difficult to adjust, it should be carefully checked and diagnosed.

For more information on specific contour problems and their diagnostics, see the section on "Contour problems - special tests, diagnostics and recommendations" in this manual.

1) Collecting and analyzing data of stable closed-loop operation under normal conditions (Automatic regulator mode).

Obligatory for data collection are normal stable circuit conditions. In this case, you can check the valve operating range and control efficiency before optimizing the circuit.

It is highly recommended to change the SP controller setting (jump) to observe the trend of the process response (load) prior to optimization. After changing the SP reference, allow the contour to stabilize so that the variability of the process can be analyzed.

Use the received data to verify that the loop operates under normal conditions:

• The controller output does not work at the end of the range (0% or 100%);

• The valve does not work near its seat;

If these conditions are not met, the valve or the final control element may need to be replaced.

• There are no cyclic oscillations in a closed loop.

If the circuit generates oscillations in the automatic mode, but is quiet in manual mode, then the cause lies in a closed loop. The cause of the oscillation may be non-linearity of the hysteresis or poor tuning.

Oscillations caused by the "sticking" of the valve have a sawtooth look, but the oscillations due to the nonlinearity of the contour may also look similar.

In a linear loop, cyclic fluctuations due to poor settings will be sinusoidal.

The period of oscillations caused by hysteresis increases when the PV variable is close to the setpoint SP, as the influence of the hysteresis increases. Analysis of the data will reveal the cause of the cyclical fluctuations in the circuit.

2) Collect and analyze stable open-loop data. Manual mode of the controller

Put the regulator in manual mode and for a certain time, collect the data of the variable PV variable.

Analyzing the data obtained in the manual mode of the regulator allows you to determine the range

noise (noise) and the variability of the circuit in this mode. In addition, analysis of the spectral power density of the signal will help to detect hidden cycles arising from previous control circuits or mechanical problems in the process scheme.

• Carefully check to see if any periodic disturbances are visible on the trend, the load jumps of the circuit. If they are, try to understand where these perturbations come from. Call up the function of frequency analysis (Power spectral density), which will help to identify the cyclical disturbances of the process. If you manage to eliminate or weaken them, your controller will work much better.

   It is not necessary to expect that the PID controller will eliminate cyclic disturbances caused by the functioning of the previous circuits, unless these cycles are slow enough compared to the speed of the loop under test. It is possible that in order to determine the source of cyclic oscillations, it will be necessary to alternately test and analyze the spectral power density of the signals of the preceding circuits, moving farther back along the technological scheme. In this case, it is necessary to search for each power spectrum peak at the oscillation frequency of the initial circuit.

• What is the noise level in the controlled variable? If the interference level exceeds 3%, installing a PV filter will improve the quality of the control.

 Since the derived component of the PID controller (D) operates on the derivative of the PV signal, any noise in the process with this component is greatly amplified. The filter will help to use D in those circuits where it was not possible before, and the correct application of the D component can significantly improve the quality of regulation.

3) Checking the loop for hysteresis

The hysteresis test is performed in the manual mode of the controller: it is necessary to make several changes (abruptly) of the CO output - two steps in one direction and one step in the other. After the test is over, the data analysis function for hysteresis Hysteresis check is started. If the hysteresis of your circuit is more than 1% for valves with positioner and 3% for valves without positioners, then to reduce the hysteresis you should consider repairing or replacing this equipment. With a hysteresis of 1% to 4%, the performance of the circuit decreases. Hysteresis of more than 3% with a rigid adjustment of the regulator causes cyclic fluctuations, similar in character to the oscillations that occur when the setting is too aggressive.

For more details, see the relevant section of the Guide below.

4) Check Valve Adhesion

To check for sticking of the valve, it is necessary to make several successive changes of CO in one direction in the manual mode of the controller: one large step and then several small steps. This test can also be performed as a continuation of the hysteresis test, namely: a series of small steps (0.5% change) is fed to the output of the CO controller in the same direction as the last step of the hysteresis test. To analyze the obtained contour data, the Stiction check function is called.

 The sticking phenomenon of the valve is very harmful, much worse than all other possible problems with the valve. For many technological processes, the adhesion value of 0.5% is too large. The presence of adhesion guarantees the presence of cyclic fluctuations and high contour variability.

5) Checking the linearity / non-linearity of the contour

To check the non-linearity of the circuit in the manual mode of the regulator operation, a series of identical step-jumps at the controller output must be made at different points of the range, for example, 15% steps at the points 5%, 20%, 35%, 50%, 65%, 80% and 95 % of the CO scale. After each step it is necessary to wait until the process calms down. A similar test can be performed with a closed loop, changing the task SP, but provided that after each step the circuit (PV and CO variables) completely stabilizes in the new state. It is also necessary to ensure that the test includes the minimum (0%) and maximum (100 %) allowable values ​​of SP.

Analysis. Using the collected data, construct a graphical characteristic of the process to determine if the process (circuit load) is nonlinear and to what extent. Find the parts of the curve with the smallest and largest slope. The amount of slope is equal to the process gain in the given section. The ratio of the maximum and minimum values ​​of the transfer coefficient of the PG process should be no more than 3x, preferably less than 2. If this ratio exceeds 3, then the output linearizer (or modify the existing one) should be included in the circuit.

Another way is to analyze the data collected at different points of the CO range using the Process Modeler process simulation function. Are models (or PID settings) different at different points in the range? If the parameters differ by more than 2 times, this contour should be linearized.

If you can not linearize the outline, you should use the most conservative, careful settings, which can be found in the Summary table of the settings of this circuit.

There are PID control loops with separation of the output signal range between two (or more) valves. Depending on the output of the CO, the controller works with one or the other valve, the switching usually takes place at 50%. For example, if CO <50%, the circuit cools the product using cold water or oil after the heat exchanger, and at CO> 50% - warms the product with steam, hot water or heated oil. Usually such circuits are very nonlinear and the output linearizer can be very useful for such a circuit.

Do not use the output linearizer to correct the nonlinearity of the pH of the circuits (pH-value of hydrogen ion concentration) - they need an input linearizer. In such circuits, the method of determining the Gain coefficient, depending on the input variable or the error (SP-PV) of the regulator, should be used.

For more details on non-linearity and construction of the output linearizer, see the corresponding section of the manual below.

6) Checking the contour asymmetry

To control the possible contour asymmetry, repeat in manual mode all or the last steps of the previous non-linearity test (see step 5 above), but in the opposite direction. Verify that the reaction of the process (load) is the same in both directions. For this you can, for example, compare PID settings calculated for one and the other direction. If asymmetry is detected, it must be eliminated or more conservative, i.e. less rapid, cautious settings should be used.

7) Comparative statistical analysis of the operation of a closed and open loop under normal conditions.

This step is optional, but very useful. In the Time plot on the trend data collected in the manual mode of the controller, select (zoom in) a section with a normally running process, that is, without unexpected surges of load or test jumps. . Perform statistical analysis of these data and note the magnitude of the variability of the process. Repeat the same with the data obtained in the automatic regulator mode, and check whether the variability in this mode has increased?

Does the regulator improve performance or performance? Are there cyclical fluctuations in the closed loop (too aggressive setting of the regulator)?

The data for this analysis can also be collected especially when the circuit is running smoothly without test jumps.

8) Determine the optimum settings for your controller

The program Xtune calculates several variants of the PID control factors P, I, D (in the Vision Framework it is PG, TI, TD) and F-time constant of the PV filter, which can be considered as the 4th loop adjustment factor. These options allow you to find an acceptable setting.

From the variants calculated by the HTune, choose the setting corresponding to the hardest or "worst" possible case of your circuit: it is either the least aggressive, that is, the slow, tuning or worst process model, that is, the model with the maximum delay time DT and the highest process transfer coefficient Gain . To verify that this is really the worst case, compare the real response to the change in the setpoint with the simulation data.

9) Analysis of contour variability with new settings

Assemble the adjustable variable PV and the output of the closed loop CO regulator in a stable state and under normal operating conditions. Analysis. Using the results of the statistical analysis before and after optimization, determine how the optimization affected the contour variability. The contour "before" and "after" can also be compared using spectral density spectra (Frequency analysis).


This manual discusses the Advanced version of ExperTune with connection to the control loops via the DDE server. The instruction is intended for software engineers and industrial technologists.

You can work with the program only if the program and its license key are installed on the workstation.

Calling from the Vision Framework application

ExperTune (XTune) is fully integrated into the ProcessSuite operator interface and is called from the Vision Framework application via a standard control loop (Fig. 1). In this case, the program's working window (the Experete faceplate, the XT faceplate) is already connected to the corresponding contour. In this environment, you can simultaneously work with only one control loop. In ExperTune there is a window-faceplate that dynamically snaps to the same tags of the Vision Framework application as the standard ProcessSuite (PS faceplate). After calling ExperTune, it is advisable to minimize the Vision window and proceed to work with the XT faceplate. The ExperTune window can be expanded or fully expanded.


Process variable PV - adjustable process variable

Setpoint SP - reference, controller setting

Controller Output CO - controller output

The panel for changing the CO output is only available in the manual mode of the controller.

Falling Menus:

Archive - enable / disable data collection

View - choose the layout option for the ExperTune window from 3 parts:

• only the faceplate,

• Faceplate + Trend

• Faceplate + Trend + Control Display Panel

Options - Set up the data representation in the window: trends, scales, colors, etc., call the report

Regardless of the Vision, the program is called up via the menu: Start-Programs-ExperTune-DDE PID Tuner or shortcut on the desktop. In this case, the Vision application acts as a DDE server. Another DDE server can be the APACS Realtime I / O Server.

Note: In this manual, only the version of the program that is designed to work with the DDE server is considered.

During the adjustment process, a loop binding file is created for each control loop to the XTouch screen (.tun path file). Work with these files is shown below in the pictures of this paragraph.

The files for connecting and operating the control loops
The initial XTune window and the contour file edit panel

Note. If you select the Options menu in the "Faceplate viewer" window - Always on top, this window will always be on the screen ahead of all the called faceplates. This is convenient if you need to call the faceplay sequentially one by one.

When the faceplate is closed, XTune remembers the contents and position of each faceplate on the screen.

In XTune, it is also possible to create contour labels that instantly call up a faceplate of the desired contour. This also provides work with a group of contours

The number of screenshots on screen is not, in principle, limited

Simulate software simulator simulate

In XTune there is a software simulator of PID control loops for the main types of technological processes

Additional trends and contours in the XT window of the faceplate

To the basic control loop (Faceplate loop), which is represented in the HT window of the faceplate with three variables: PV, SP and CO, you can add additional trends for a number of other variables ("Extra_Trend"), as well as additional contours (the object "Additional_contours"). Further, these objects are displayed, tested, their data is written to the archive and analyzed together with the main contour. There are many examples where you need to monitor other processes when testing the outline for configuration and analysis. This, for example, cascade control, interdependent circuits, proactive compensation regulation, etc.

For each contour connected to the XTune (ie, in each .tun file), you can create up to 64 additional trends. But only the trends you need are displayed on the display in the faceplate window.

In addition, from variables for which additional trends are defined, you can create many additional circuits for configuration and analysis. 

Configuring additional trends for the CT faceplate

 In the Edit Main Setup (.tun file) Edit Setup window, click the "Advanced" button. Then, on the panel that appears, click "Extra Trends" to display the Setup Extra Trends additional detection and configuration panel.

At the initial call, this panel is empty. To set the first Additional_Trend, and then the following trends, you must click the Add Trends button. Each Additional_Trend needs to be given a unique Name, enter the DDE information of the connection with the variable, information about the scale, and also the parameters of the trend indication in the HT window of the faceplate. The parameters of the trend display can be set directly through the faceplate.

The PV, CO, and SP names represent the variables of the main loop and can not be used for AdditionalTrends. After the trend configuration is completed, you need to click the Test button to verify that the trend is correctly identified and linked.

   All additional trends are archived on the disk and then they can be used for configuration and analysis in a manner similar to the trends of the main circuit.

 If the contour with additional trends has archived .tun files, then you can no longer delete or rename existing trends or change their DDE communication addresses with XTune. This limitation allows you to preserve the integrity of the data in the archive.

If you want to change existing trends, you must either delete all the archived files of the current path, or create a new path using the SaveAs button in the Edit Setup window.

Creating an additional path

Each contour name (.tun file) in the list of the initial window of ExperTune represents a number of archive files that contain data for both the primary loop and all additional lines and additional circuits configured for this loop.

To create the main loop, use the Edit Setup edit box, but the Extension loops are configured using the Extra Loops Setup special window. To open this panel, click the "Advanced" button in the Edit Setup window or select the Options menu in the Offline file window and then Loop Setup.

To create any additional contours, you must assign its variables: PV (adjustable variable), CO (controller output) and SP (Task), where PV and CO are mandatory, and SP may be absent. In this case, the names of the real variables that perform the role of PV, CO, SP are taken from the Additional_Trends list. In addition, for each additional contour, the name of the loop, the type of regulator and the direction of the reaction of the process to the action of the regulator are indicated: direct or inverse.

Thus, the Additional circuit is a named pair or three variables of PV-CO-SP.

In the configuration window Extra Loops Setup, select the desired contour in the "falling" list at the top of the window. Here you can add or remove additional_contours.

The main loop (Faceplate loop), which is specified in the name of the HT window of the faceplate, can also be selected, but can not be changed. You can make changes to this path only in the Edit Setup window.

Enter the new name of the Extension, then select and mark with the "label" two or three variables that form this contour as PV, CO, and SP.

Working with cascade control loop

In the cascade circuit, the output of one (primary PRI) regulator serves as the reference for another (secondary SEC) regulator. In order to avoid interaction in the cascade, the mutual influence of the two circuits, the outer (primary) contour must be at least 3 times slower than the secondary circuit.

The speed of the loop depends on the setting parameter of the controller I (TI) -constant integration time and is characterized by the indicator RRT (Relative reaction time and loop), which is calculated in the contour analysis. The RRT of the external loop must be 3 times greater than the RRT of the internal (secondary) circuit. If this requirement is not met, then it is necessary to reconfigure the outer loop so that its RRT is 3 times larger than that of the inner loop. In addition to RRT, the speed of contours can be estimated from the values ​​of the settings of I (TI).

To configure the cascade, first configure the internal loop in local mode (with the cascade open). Then go into cascade mode and configure the outer loop.

Procedure for testing and setting up cascading circuits:

• Set the internal (secondary) circuit to the local LOCAL mode (the cascade is open, the SP job is entered manually).

• Test the internal circuit and collect the PV and CO data, as described in the corresponding section of the Instruction.

• Load the settings calculated by XTune into the secondary loop PID controller.

• Set the secondary (internal) circuit to cascade mode, in which the input of the reference / setpoint SP of the internal circuit is connected to the output of the regulator of the external (primary) circuit.

• Test and collect the PV and CO primary data. The controller output serves as the reference for the secondary circuit.

• Compare the values ​​of the integration time I (TI) obtained for the primary and secondary circuits. The value for the external (primary) circuit in units of time / repeat (min / repeat) should be 3-4 times greater than the secondary circuit. If not, increase the primary circuit integration time so that the required ratio is satisfied.

The program XTune has a software simulator cascade, which will help users to work with difficult cascaded circuits. Using the software simulator, you can configure both contours of the cascade simultaneously.

Testing and setting up two cascade circuits with one test

This possibility is based on the use of the program simulator ExperTune.

The secondary (internal SEC) circuit is tested and modeled in the usual way: PV data and the output of the CO controller are collected. However, for the external (primary PRI) circuit, the process model is found by using the secondary loop instead of the primary regulator output.

Testing the cascade "jump" can be performed both in the primary and secondary circuits. It should be ensured that the collection of data begins and ends when the regulated variables of the PV of both stages are stable.

Example procedure:

• Connect ExperTune to the secondary loop of the cascade and create an additional trend for the PV variable of the primary (external) loop.

• Perform the test by applying a test "jump" to the secondary circuit. Collect and record in the secondary data archive (PV, CO) and PV variable of the primary (external) circuit. In this case, the primary circuit must function normally, and its adjustable PV at the beginning and at the end of the data collection must be stable.

• Create an additional contour of two variables: PV primary and secondary PV, where the PV of the secondary circuit replaces the output of the primary loop controller. Name the resulting additional contour "Master Process" (Primary process).

Perform (Tune) the calculation of the settings for the Master Process loop, then click the Analysis button to identify the load process of the primary (external) contour, that is, to build its mathematical model. The secondary (internal) circuit is the load for the primary circuit controller.

• In the next step, select the previously obtained secondary loop test file in the archive and call the calculation of the settings for it (Tune), and then perform the analysis. After analyzing the process model "Process Model" in the Model type list, select "Start Simulator With This Model". The XT simulator of the control loop and its adjustment window are called up.

• In the window of the XTune simulator, click the "Cascade Loop" tab and enter the primary (external) process model obtained in the previous step. Then open the "Controller" tab and select the appropriate controller for the primary process. Now the program model of the cascade is obtained in the simulator.

• Next, select the secondary circuit in the simulator window and calculate the PID Tuning settings. The loop simulator will automatically calculate the settings of your secondary loop controller for the specified combination of the process model and the controller. Typically, for a secondary circuit with a safety factor of 1, do not use the "load tuning" settings, which are primarily intended for testing load disturbances. It is recommended to establish a coefficient of stability of at least 2.5 for the secondary circuit.

• Click the "Master Controller" tab and call "PID Tuning" to set the primary loop. Then, in the presented table, select the desired configuration option.

• Review the variants of the reaction of variables and the stability curves of the cascade obtained during the analysis. Try to explore on the resulting model possible options on the principle of "What will happen to the cascade, if ...".

Cascade setup example

4. Testing and data collection for analysis, diagnostics and adjustment of the control loop

Requirements for data collection during testing

Requirements for optimal PID settings, reliable modeling and analysis circuit

1. The data collection of the variable PV variable and the output of the CO regulator must start and end with a stable (quiet) state of the circuit (and process) according to the principle:

Stable state - rapid change (PV or CO jump) - a new stable state.

Particularly important is the initial phase of data collection in a calm state (before the jump).

End testing can be within 5% of the stable state

Testing involves a jump in PV or CO. In the automatic mode of the regulator, during the testing, the SP (setpoint) setting changes, in the manual mode, the output of the CO controller.

Stable state means that the regulated variable PV and the output of the CO regulator are stable simultaneously, their trends are horizontal lines that can deviate from the straight line no more than within the "normal" noise of the control object. In this case, not necessarily PV = SP.

If significant changes in PV, CO or cyclic fluctuations occur, then this is not a stable state. Random, unreasonable disturbances, bursts of load, too, do not provide "good" data for tuning and analysis.

The normal noise for a given circuit is a small change, a disturbance of the adjustable PV variable, not caused by the regulation itself. This "noise" is due to electrical interference, electromagnetic interference, turbulence in the flow control circuits, due to possible waves in tanks with a liquid product, and the like.

2. During testing, the load of the circuit should not change, there should be no random disturbances, abnormal surges of the controlled variable PV, otherwise the data obtained will be bad. With such disturbances, the PV starts to change before the regulator goes out, and the program will incorrectly evaluate these data as a process with an infinite transmission coefficient. The test data scale should be as linear as possible. For example, giving a pneumatic signal to the valve to create a "jump" will not give good data for adjusting the regulator, since there will not be a jump in them. Instead, change the SP or put the circuit in manual mode and give a jump at the CO output.

In the case of "noisy" circuits, the collected test data, that is, changes in the PV variable caused by the test, should be about 10 times greater than the maximum amplitude of the noise fluctuation (peak to peak). The same applies to changes in the output of the CO regulator.

What is the disturbance / load jump Load (Process) Upset?

This change in the process which is not caused by a change in the setting of the SP regulator, an external effect that causes the adjustable PV variable to deviate from the SP setting in conditions of stable normal contour operation. The sign of load disturbance is that the PV starts to change earlier than the CO (see the figure below). Examples:

Flow control loop - change of pressure lower in the process chain.

Level regulation by output flow - increasing the input flow will be a load disturbance.

Another example: steam for heating is injected into the cold water stream, the result is a change in the temperature of the cold water at the input of the apparatus.


3. If your circuit is unstable or generates cyclical fluctuations, try the following:

• Move the loop to the manual mode of the regulator and wait until it settles down

• If the circuit can not be switched to manual mode or in this mode it drifts, then in automatic mode, enter a low PG (usually about 0.2) and wait for the circuit to stabilize.

4. For better operation of ExperTune, filtering of the adjustable PV variable should be avoided. If the filter is used in the system, then the XTune program will filter at your request, provided that the option Un-filter is selected,

5. Interval of data collection during testing. ExperTune analyzes pairs of PV and CO values ​​read at a certain frequency. Data collection should be performed at an interval equal to the controller operating cycle or 4-10 times lower than the DeadTime-equivalent time lag (dead zone) of the controlled process. This interval is measured in seconds and is also used to update trends and diagrams.

It is usually considered optimal when the reading interval is equal to the controller cycle in the controller. The interval (controller cycle) on one side should be short enough compared to Dead Time (DT), so as not to introduce additional delay into the loop, but on the other hand, not too small to not overload the controller. The DT value is the time required by the PV to start changing after changing the output of the controller.

In the APACS + / QUADLOG system, the controller cycle is equal to the scan cycle of the controller.

It is recommended that the data collection interval is 4-10 times less than the time delay of the process, otherwise the quality of analysis and adjustment is reduced. For example, if the data reading interval is close to DT, then, by reducing it only 2 times, you can double the efficiency of the circuit. The program defaults to make this interval equal to 0.1 Dead Time. If the interval does not meet the requirements, the program will issue a warning message.

Data compression. The XTune program uses for analysis and setting a maximum of 1025 pairs of values

data. If the collected test data contains more than 1025 pairs, then the sampled sample is compressed to 1025 pairs. The user can assign a different limit. In principle, the program can read up to 1 million points, but it takes several minutes to read large amounts of data.

A very high quality of tuning can be achieved even with a data volume of 200 to 500 pairs of values, and there is no need to use large samples.

Typical procedures for testing PID control loops

1. Regulator in manual mode (open loop):

a) Check that the output of the CO regulator is not in the 0% or 100% state or in the other saturation limit state. Otherwise, put the output in the intermediate state in the range between 5% and 95% (or output it from saturation). The valve in the limiting state usually becomes a nonlinear element.

b) Wait until the contour is calmed down and goes into a stable state. Enable data logging (archiving).

c) Quickly (abruptly) change the regulator output by approximately 10%

d) Wait until the adjustable variable changes (works) to a sufficient value, then return the CO output to its previous state. This step can be skipped if

It is known that your technological object reacts well to the change in the set point.

e) Wait until the contour is calmed down and goes into a stable state. Turn off the data logging.

Change in CO output in manual mode

Change of the reference (setpoint) SP in the automatic operating mode of the controller

2. Regulator in automatic mode (closed loop):

This test is similar to the previous test, but instead of jumping, the setting (setpoint) of the SP controller changes abruptly. (Figure 3)

The test is usually performed faster if you remove the integral component of the PID controller. For this, in the controller settings, the integration time constant TI must be made as large as possible, i.e. TI = 4000.

3. Regulator in automatic mode and quick jump manually at the CO output:

a) Check that the output of the CO controller is not in the 0% or 100% state or in the other saturation limit state. Otherwise, transfer the output to the intermediate state in the range between 5% and 95% (or output it from the limit state). The valve in the limiting state usually becomes a nonlinear element.

b) Wait until the contour is calmed down and goes into a stable state. Enable data logging or archiving.

c) Set the regulator to manual mode and quickly (jumpwise) change the output of the regulator by 5-10%

d) Immediately return the regulator to automatic mode

e) Wait until the contour is calmed down and goes into a stable state. Turn off the data logging.

4. Quick test for slow circuits - the regulator in manual mode:

This test is performed by applying a double pulse to the output of the regulator, which greatly speeds up the collection of data on the controlled variable, especially for slow contours with a long delay. A stopwatch or a timer is required for testing.

See Fig. 1 and 2.

a) Check that the output of the CO controller is not in the 0% or 100% state or in the other saturation limit state. Otherwise, transfer the output to the intermediate state in the range between 5% and 95% (or output it from the limit state). The valve in the limiting state usually becomes a nonlinear element.

b) Wait until the contour is calmed down and goes into a stable state. Enable data logging or archiving.

c) Set the regulator to manual mode and prepare a stopwatch or timer

d) Quickly change the output of the regulator by 10%. Turn on the stopwatch.

e) Wait until the adjustable variable changes (works) by an amount much higher than the noise level in the circuit, and then jumpwise change the regulator output in the opposite direction and by a value twice as large as in the previous (d) step, i.e. on 20%. Also, record the time T that has elapsed since the first jump, and then turn on the stopwatch again.

e) As soon as the time interval T has elapsed, return the regulator output to the initial state, which was in step b (Fig. 1).

g) Wait for 2T and then stop collecting data.

Quick test for slow contours. Regulator in manual mode.


Ручное (полуавтоматическое) тестирование


Some diagnostic messages when collecting contour data
Keys and functions of the trend window for source data



ExperTune uses exactly the data shown in the "Time data for:" window to calculate the settings and analyze the loop. If the edit command Zoom, Edit line, Average, or Undo changes the trend in this window, then the program, using the mathematical model of the process, automatically recalculates all PID settings and contour characteristics for the new original data.

The path name of the data file used is written at the top of the window title. Trends window of source data can be moved to any place on the screen. Closing this window is similar to pressing the Done tuning key and causes the end of the analysis and calculation session of the contour settings.

Editing the collected data

Initially, the collected test data (test sample) is represented in the Trend of Data window with a real trend graph, which can then be edited by the user. Editing is aimed at improving the "quality" of the raw data for subsequent calculations and may include:
• Selection of a part of the trend with "good" data (Zoom in)
• Averaging some data (Average)
• Smooth the individual sections of the trend to eliminate, for example, random peaks and noise.
The sequence of trend points can be replaced by a straight line segment (Edit0
• Elimination of the influence of the filter PV (Un-filter), etc.

Use the Zoom in function to select in the Trends window and then increase the smallest area with test data that:
1) Begins with a stable state of the circuit. (Very important)
2) Ends within 5% of the stable state.
To edit the data, use the Average and Edit Edit functions

Zoom In (Select and enlarge)

Place the mouse pointer over the first point of the data section that you want to use, and click the left key. Then select the last desired data point with the mouse and click the left key again. In this way, a section of data trends for the new window will be defined on the time scale. After that, clicking near the left or right edge of the selected area, you can stretch it accordingly to the left or right.

Click Zoom now to finally select the highlighted data and get a new window with the data trends. The number of points used for analysis should be at least 33 and not more than 1 billion.

Zoom Out - Return the full original image

Zoom Back to Previous - cancel the result of the previous Zoom In command (you can go back one step at a time)

Auto Zoom - an attempt to automatically select the best data for calculating settings

Average-Averaging part of the data in the Trends window. The function is called from the Edit menu

WARNING: Averaging changes the collected real object data.

The selection of the data for averaging is done using the cursor similar to the selection for the Zoom In function. To cancel, use the Undo option or the Cancel key.

Line Edit

This function allows you to replace a sequence of points (a section of a trend) with a straight line or a broken line, which is useful for removing noise peaks and other random data. You can even edit a single point.

To select and edit data, the "rubber band" function is used.

Data collection for the analysis of stable contour operation

For a number of analysis tasks, such as statistical and frequency analysis of the circuit, information is needed on stable normal contour operation without jumps in SP and CO. The data collection for this case differs from the tests for calculating the settings. The necessary data can be obtained in the following way:

• If the contour is stable, collect the data anew for analysis, using a semi-automatic procedure with manual data collection on and off.

• Using the Zoom In option, edit the Trend File information of the archive file of one of the previous tests of this circuit, selecting the necessary piece of data for stable contour analysis for analysis.

Examples of testing data at industrial facilities

The figures show real test data, as well as some examples of selecting a trend section or editing the collected data in order to obtain "good" data for calculating regulator settings and analysis.

Example 1. Trends of collected data have cycles and random peaks of interference.


Recommendations. Before using this data to calculate the settings, you need:

1. Select a trend line of data, highlighted in dark, using the Zoom in function.

2. Use the edit to remove the noise peaks.

3. It may be worth analyzing the spectral power density of the signal to find the cause of the cyclic oscillations.

Example 2. The process reacts to the regulating effect in one direction more quickly than in the other - the asymmetry of the process.


Recommendations. Select in turn (Zoom in) the two dark areas of data shown in the figure and get for each value of the controller settings. Use a more conservative option settings. To calculate the most conservative settings, use the Loop summary table.


The procedure for adjusting the controller

STEP 1. Connect the XTune program to the control loop via the DDE server. When you call XTune from a PS faceplate in the Vision Framework application, this connection is automatic.

STEP 2. Testing and collecting contour data:

• to enable the automatic procedure, you must click on the HT faceplate key AutoTune Sequence (Automatic setting)

• When using a semi-automatic procedure, the data collection is manually turned on and off (on the HT faceplate).

After collecting information by clicking the "Tune from archived data" button, you can call up a list of the archive test files of this circuit. Select the desired file and click the "Tune" button (Calculate controller settings).

After finishing the automatic setup procedure or after selecting the archive data file and pressing the Tune key, the Trend Data window "Time data for ..." appears with the trend data saved in the archive, and the current settings and new values ​​of the controller settings are calculated on the screen, calculated on these data, including the filter time F.

STEP 3: View and, if necessary, edit the test data presented in the Trend Box that ExperTune will use to calculate the settings.

To edit the data, the functions Zoom in, Edit, Average and Filter are used. In addition, on the faceplate in the "Controller tuning" area (Control panel tuning panel), you can select the desired option settings - with a focus on the best workaround:

• load tuning load disturbances (3 options: fastest fastest, simple fast fast or relatively slow Low)

• Setpoint tuning or its version of Lambda tuning.

Any changes made to the data in the Trend window automatically cause the XTune program to recalculate and display new PID settings. Thus, the data in the Trend Box is always the source information used by ExperTune to analyze the loop and calculate the controller settings.

STEP 4: Download new settings to the controller by clicking the Download button.

About the method of calculation of PID settings by the program ExperTune

To determine the best mathematical model of the process and calculate the PID settings, ExperTune uses a frequency analysis method and an expert system.

First, the program converts the original data presented in the Trend window into the frequency response of the process (AFC and PFC). That is why the original data (collected during testing and, possibly, edited) should start and end with a stable state of the contour. Thanks to this characteristic, ExperTune can adjust the circuit based on the results of only one stepping jump or pulse in the automatic or manual modes of the controller.

The frequency characteristic of the technological process uniquely represents and identifies this process. The quality of the characteristic and the PID settings calculated on its basis depend on the quality of the original primary data. The random disturbance of the load, reflected in these data, distorts the frequency response. Let, for example, you use such data. But with a similar process disturbance, the controlled variable PV changes earlier than the output of the CO regulator. Therefore, the gain of the process Gain on the AFC will tend to an infinitely large value, since PV has changed, and the change in the CO output is zero.

Then ExperTune launches an expert system that finds the best settings and model of the process. 8-10 options (categories) of PI and PID settings are calculated, including PV filters.

It is often asked whether ExperTune uses the Ziegler-Nichols method? Our quick tuning option "Load tuning-fast" is closest to Ziegler-Nichols, but the fastest (optimal) version of "Load tuning-fastest" is better than Ziegler-Nichols. ExperTune recommends using this setting.

Regulator setting

Controller Tuning Panel

The controller setting panel (highlighted with a frame on the faceplate) opens with the call of the PID calculation function: from the faceplate or Trends window with the "Tune" key; via Options menu in these windows; when closing the Trend window.

The program ExperTune calculates the settings of the controller P, I, D and the filter time F from the initial data presented in the Trend window "Time data for: ..." with additional parameters / requirements in the Controller Tuning.

Initially, the program offers its own standard version with the default parameters. Usually this is the option shown in the figure: "Load tuning - Fastest". The values ​​of the new recommended settings are presented in the New column on the faceplate, and the additional parameters used are displayed on the Controller's Control Panel.

In general, this panel selects the option (category) of the settings and specifies additional requirements and parameters that ultimately determine, together with the test data, the type of regulation and the values ​​of the recommended settings calculated by ExperTune.

Possible choices include:

• Type of control algorithm - PI or PID, that is, the use of a differential component (derivative control)

• Factor of stability of the contour Safety factor

• Variants of settings with a focus on better compensation of Load Tuning with different speed or optimal response when changing the job SP (Setpoint tuning)

• Using the PV filter and its type

The program ExperTune finds several variants of settings, optimal for different types of regulation and their applications. All these categories of settings (10 variants) are presented in the Table of recommended settings of the "PID Tuning Grid", which is called by the PID Grid key.

The desired configuration is selected in the "drop-down list" in the box under the PID Grid key, and in the Table itself the selected option is highlighted with a frame. Explain the options in the next paragraph.

The values ​​of the adjusting parameters of the controller P, I, D, F, calculated and selected in accordance with the requirements of the request for the Controller Control Panel, are shown on the faceplate in the New column to the left of this panel.

If you change the request in the Settings Panel, as well as when changing the initial data in the Trend window, the settings are recalculated and updated.


Safety Factor (SF) - Stability, stability, coefficient, determining the degree of stability of the contour to changes in the dynamic characteristics of the load / process. SF can take values ​​at least 1, by default SF = 2.5.

Use derivative if possible - use the differential component (derivative control) if possible.

If this is not necessary or harmful for the process, then the zero value of the differentiation time constant D = 0 is set.

Find PV Filter - find the PV filter. The type of filter is selected in the "falling" list. For more information about the PV filter see the special section of the manual.

Un-filter the PV - remove the action of the PV filter.

Loop Summary-A summary table (log) of the contour settings.

Add to Table - add the setting to the Pivot Table (log).

If the parameter is not available or is not used in this category of settings, then N / A is indicated.

Done tuning - End of regulator setting

Download - upload new settings to the controller.

Probable Performance Increase - Probable increase of efficiency / productivity of a contour

Comment. The filter time constant F can be considered as the fourth tuning parameter, and talk about PIDF settings.

When to use the differential control component.

The use of derivative control (D) allows for a greater use of proportional (P) and especially integral (I) effects, which can result in a much quicker loop response to perturbations (see figure) With proper application, this is particularly evident in case second-order processes, such as temperature control, but can also improve the transient response and most other circuits.

D must be used with caution. If D is too much, this makes the circuit unstable, if too little, it does not work, but can cause the valve to "jitter", increasing its wear. The ExperTune program ensures optimal use of the differential component D.


• Do not use D with processes that have almost all the latency - pure dead time delay

• Carefully use D in circuits with "noise", since D can increase interference. The PV filter can help.

• Always consider the possibility of increasing valve wear due to D and use compromise solutions.

• If the controller does not restrict the action of the derivative, then D can not be applied

Example of application of the differential component D: blue-PI, red-PID


Loop Performance - the effectiveness of the application of the control loop in the process, affecting the quality and cost of production. It is determined by the speed and quality of regulation in the circuit. Ultimately, this indicator is proportional to the economic efficiency of the circuit.

Performance Increase - Improve the performance / efficiency of the circuit

This indicator (PP) depends on the speed of the loop, the speed and accuracy of the control and allows you to estimate how much better the PID controller with the new PID settings responds to the disturbances of the load (process). It is used for comparative analysis of contour settings.

If the contour is optimally tuned, it has minimal variability and can work closer to the set task, in accordance with the regulations and with less loss of resources.

The PP is an estimate of the probable relative (in%) increase in control efficiency under load disturbances, calculated from the integral absolute control error IAE, assuming that there is no over-regulation due to load disturbances both under "old" and "new" settings.

In the case of overshoot, the PP becomes a "rough" indicator, approximately a proponic improvement in the IAE integral error. Nevertheless, even so, it remains a sufficiently good tool for measuring and predicting the "productivity" of the circuit.

Usually this indicator is directly proportional to the amount of money that can be saved by using the new settings. With a "bad" setup, accidental disturbance of the load in one direction can lead to a violation of the production regulations and a decrease in product quality. Changing the same load in the other direction will require additional costs of expensive materials or energy to keep production within the technological boundaries. Therefore, the optimal tuning will ensure the economy of resources at the required quality of the product.

For example, adding an MTBE additive to gasoline increases the octane number. But it's an expensive additive, and you want to add just as much as necessary to get the right octane number. Add less prohibits regulations. Therefore, it is necessary to adjust the consumption of the additive as close as possible to the task. This feature is provided by the regulator setting, which provides the minimum value of the accumulated (total) absolute error IAE.

Find PV Filter - find the PV filter of the specified type

The type of filter is selected in the "falling" list.

The program will calculate the largest possible filter of the specified type (filter time F) for the selected PID setting so that it does not significantly reduce the loop efficiency.

Via the Edit Setup window for editing the connection parameters of the loop to ExperTune, you can set the filter time unit: Edit Setup - Advanced - Loop Setup - Advanced. If you select Same as D (the same as D), then the filter unit will be the same as the D time constant of the controller.

 You can not only use the recommended filter, but also try to enter your filter time - current or new. Trends in the transition characteristics in the Control Loop Simulation - Setpoint (- Load upset) and Robustness Plots contours are instantly updated in accordance with the new type and size of the filter. In the Control Loop Simulation -Measurement noise response window, it is also possible to analyze the effect of the filter on interference suppression and reducing valve wear.

In the PID Grid options window (PIDF Table), all the filters recommended for each category of settings are presented.

IMPORTANT NOTICE. The initial data for ExperTune, used for modeling, analysis and calculation of regulator settings, should be collected without a filter, otherwise all data will be distorted.

The filter can be removed automatically using the special option Un-filter the PV (Remove the PV filter).

For more information about the PV filter, see the PV Filter section in the control loop.

Un-filter the PV (Cancel PV filter)

This effective function allows the user to see and use raw raw PV data prior to filtering. Thanks to this option, even filtered data can be collected for analysis, and then instruct XTune to remove the filter. For the accurate and correct modeling, analysis and calculation of PID settings, the program needs "clean" data without filtering. But sometimes it is more convenient to compile and archive PV data with a filter. Then, ExperTune automatically removes the filter and presents the user in the Trends window with raw data without filtering, providing conditions for more accurate calculations.

The type of filter used by the Un-filter function is displayed on the faceplate in the window with the "falling" list of "PV Filter". This can be one of the 4 types discussed in this section of the Guide.

In order for the filter to turn off automatically, it is necessary to select the "Un-filter the PV" function on the XT faceplate in the Controller Tuning Control Panel area above the Loop Summary key. You can also use the same option in the Edit menu or the key in the Trend Trends window.

  The current value of the filter time F is shown in the column of the current 4 settings of the controller "Current P, I, D, F". If the current value of F is read from the controller, it is displayed against a gray background. The filter cancellation function uses the filter time value that was at the time of data archiving, which may differ from the current value.

The filter time constant used by the Un-filter PV function can be seen by placing the cursor (without a click) on the same key in the Trend window.

Comment. The "Find PV Filter" option is a calculation of the recommended filter value, and the "Un-filter PV" option is a completely different, independent function.

Table of recommended options for PID settings- PID Tuning Grid

This table shows all the recommended options for the contour settings calculated by the program from the initial data. The trend window for PI and PID control in 5 typical categories with different requirements for speed and accuracy of control and contour stability.

The loop configuration parameters include 3 PI and PID control parameters and the PV filter, which are calculated in units received in the controller or system to which the XTune program is connected:

P - factor of proportionality or gain (in the system APACS + / QUADLOG is denoted by PG),

I is the integration time constant in time / repeat time units or inverse units (in APACS + / QUADLOG it is TI in minutes),

D is the time constant of differentiation in the same units of time as the integral coefficient I (in APACS + / QUADLOG this is TD in minutes).

Filter (F) - the 4th setting parameter is the time constant of the PV filter. The unit of time for the filter can be set in the Edit Setup-Advanced-Loop Setup window. By default, this unit is taken equal to the unit of time for differentiation D.

Each configuration option is additionally accompanied by the value of the parameter RRT (relative loop response time in seconds), which characterizes the loop speed.

If some parameter is not used in this category of settings or is unavailable, then its value is set to N / A.

In this window, the user can also change several common parameters: the stability factor of the Safety Factor contour, the type of the PV filter, etc. When the window is closed, the settings recalculated after these changes are shown in the "New" column.

Terms and Definitions

Conservative regulation, conservative settings-careful, "soft", relatively slow regulation with orientation to ensure the stability of the contour.

Aggressive, hard settings, regulation - focus on rapid control and maximize loop performance

Options by category

Load tuning-settings "For load", provide for most circuits optimal control and compensation of disturbances, load jumps of the regulator (process). For this category, the program calculates 3 variants that differ in the degree of aggressiveness of regulation:

• Fastest - fastest

• Fast - fast

• Slow - slow

Setpoint tuning- settings "To work out the task" are intended primarily for optimal control of the process when the SP

Lambda tuning- "lambda" settings, a variety of settings "for working out the task"

Settings for better load disturbance compensation

For most circuits, ExperTune recommends using PI or PID control at the fastest "Load" setting, that is, Load tuning - Fastest. This is the most difficult for regulation and at the same time the most common, often occurring case, in particular integrating processes and lag-processes with tightening.

Example of load disturbance. Level regulation in the tank with a control valve at the outlet. In this case, changing the flow rate at the input is a disturbing change in the load of the regulator

A sign of disturbance of the load of the regulator (process) can be the appearance of an apparently unreasonable change in the controlled variable PV in the absence of changes in the setting of the SP or in the output of the regulator CO, the change in the PV begins earlier than the change in CO.

In practice, when choosing settings, there is always a trade-off between the speed of the circuit and its sensitivity to changes in the dynamic parameters of the regulated process (proportionality Gain and dead time Dead time). Usually the fastest option of settings with a low coefficient of stability is simultaneously the most sensitive to changes in the dynamics of the technological process. Conversely, the circuit with the slowest settings and the highest degree of stability is the least sensitive, that is, it is more stable.

An example of a loop response to a change in the SP job before and after the "Load fastest" setting

The influence of the factor of stability (Safety factor) on the settings "For load" - category Load tuning

The SF factor allows the user to control the stability of the contour in the settings of this category. In other categories: "Setpoint tuning" and "Lambda tuning" this role is played by the time constant of the transition characteristic of the Response (lambda) time loop.

The default degree of stability SF = 2.5 gives a conservative setting that meets the requirements of the technology in most practical cases. To get a faster loop response, enter SF <2.5. At SF = 1, the fastest reaction is obtained, but without a margin of stability.

With the stability factor Safety factor = 1, the following options for the "For load" settings are obtained:

Option Control characteristic at SF = 1

Fastest (fastest): The best option is the minimum absolute error in the load jump

Fast: Quarter is amplitude damping of the oscillations (the amplitude of each next half-period of the transient response is approximately half the amplitude of the previous half-period)

Slow (slow): Redundancy of 10%

With SF = 1, the settings are very sensitive to small changes in the dynamics of the process (gain and deadtime delay). It must be borne in mind that most of the control loops are partly non-linear. Therefore, in general, for the stability of the circuit, compensation of nonlinearity, and also to reduce overshoot, the coefficient SF> 1 and quite conservative settings.

The problem of overshoot

The overshoot (the PV output for the SP setting when the loop reacts to the load or job jumps) depends on many factors, including the dynamic characteristics and nonlinearities of the process, and the cause of the change is the load or the SP reference.

In general, in order to eliminate overshoot at the "For Load" settings, it is necessary to reduce the integral action of the regulator by 3 times. For example, if the integration time constant I (TI) is measured in minutes / repeat, multiply I by 3, and if in inverse units - divide by 3.

The use of the option "Setpoint / lambda tuning" usually eliminates overshoot due to load disturbances.

Settings that provide the optimal adjustment of the job changes - Setpoint tuning

This option is also called "Lambda tuning", that is, these two terms are synonyms. In the Options Tables window, under the name "Lambda tuning", an additional variant of the settings "To work on the job" is given.

In this category, instead of the Safety Factor stability factor, the parameter Response time is set in seconds-the time constant of the first-order PV transition characteristic when the SP job is changed. This parameter determines the speed of the contour when the new task is being developed and also the degree of its stability. For a faster reaction, a smaller value of the time constant should be given, and for a slower reaction, a larger value.

The settings of this category are calculated so as to ensure the operation of a closed loop with the total reaction time, which is composed of the time found by the program, the dead time and the specified response time (Response time). The response of the loop to the change in SP is first delayed for a time Dead time.

For most circuits, ExperTune recommends the Load Tuning - Fastest setting for PI and PID control, but if you want to eliminate the overshoot option, use the Setpoint (Lambda) tuning settings for the job. Therefore, this type of setting is popular in a number of industries, where operators - technologists do not want to work with overshoot.

By default, the program sets a "conservative", a sufficiently large value of the reaction time, to ensure that there is no overshoot when the SP job is changed. To get the most "dense" and quick workout in this category of settings, reduce the initial value of the time parameter by 3 times.

Disadvantages of the category Setpoint tuning:

• There are no settings for circuits with an integrating link

• There are no PI settings for regulating second-order processes, although for most such PID processes the settings in this category exist

For integrating circuits, you can apply the Lambda tuning option - see the next paragraph for this.

The standard "Lambda" setting is Lambda tuning

This option also refers to the category of settings that focus on the optimal working of the SP job without overshooting, and is consistent with the practical tuning methods used by specialists familiar with the "lambda" regulatory method.

At the bottom of the settings column is a small "Lambda" field where the lambda type of the controlled process is selected:

1. Lag rule - 1st order process with gain Gain and pure delay Dead time

2. Intg rule - integrating process in conjunction with the Dead time delay

If these characteristics are not applicable to your process, then they will not be accessed in the window.

When Lag rule is selected, this option is the same as Setpoint tuning, except that the default (default) Lambda time is usually greater than Setpoint Response time.

Quality of Frequency Data Fit - an estimation of quality of the initial data used for calculation of adjustments.

The optimality and efficiency of the settings calculated by the program depends on the quality of the original data presented in the Trend window.

This indicator gives an overall assessment of how much smoothed, free of interference (noise), correctly collected and reliable are your data on the work of the circuit. The indicator can take the following values:

excellent - excellent very good - very good good - good fair - satisfactory poor - weak, not enough

questionable - doubtfully very questionable - very doubtful

The questionable quality of the data means that the received PID settings are most likely unreliable and should not be used.

ExperTune evaluates the quality of the input data for the amplitude (frequency response) and phase (PFC) frequency characteristics of your process (Bode graphs), and this estimate shows how closely the mathematical model found by the program coincides with the real frequency response.

Frequency response, in which both curves: AFC and PFC - amplitude and phase, smoothly decrease with increasing frequency, gives a better indicator of data quality. The better the source data, the better and frequency response.

How to improve the quality of data

• Check the data in the Trend window for compliance with all requirements.

Data collection for both PV and CO should start and end with a stable contour state, first you must first collect a sufficient amount of stable data. For details, see the relevant section of this manual.

• Verify that the collected data is not the result of two different tests with an extra jump in the SP or CO output.

• Sometimes you can improve data by editing them in the Trend window: select a trend section with "good" data; remove random peaks; smooth out excessive noise, remove the "bad" part of the trends (see Zoom In and other editing options)

• If the load (process) is disturbed during testing, then it is necessary to collect all the data again taking into account all requirements.

• Do not use the PV filter to improve the quality of the data - this does not help. Adding a filter to the loop will lower the amplitude coefficient beyond the filter time, but it will not reduce the fluctuation of the curve. If the filter is large enough, then the program, in addition to the process, will also try to tune in to the filter

• There are contours that do not produce high-quality data at all

Even very good data quality does not guarantee that optimal and qualitative PID settings will be obtained. It is always necessary to consider the following questions: 1) Do these raw data represent the full operating range and working conditions of the circuit? 2) Do these data belong to the same contour that you are setting up?

Sample interval - The interval (period) in seconds of the cyclic operation of the controller and calculation of the output signal by the PID algorithm.

The cycle (period) of the regulator should be small enough not to introduce a noticeable additional delay of the "dead time" type into the loop, but not enough to cause the controller to overload or reduce the accuracy of the PID algorithm calculations.

For optimal operation, the controller cycle must be 4-10 times shorter than the process dead time.

The term Sample interval is also used to indicate the interval of data collection by the program ExperTune. In order to obtain reliable data and build a sufficiently accurate process model for analysis and calculation of PID settings, this interval should also be 4-10 times less than the delay time of the process. Therefore, it is best, optimally, when the data collection interval for ExperTune is equal to the cycle time of the controller (controller). With the same interval, the variable values, trends and other dynamic images are updated on the faceplate and other ExperTune windows.

The program checks the controller / controller operating period and the data collection interval for compliance. If the requirements are met, the message Sample interval checks OK appears in the Tables of options settings window at the bottom.

If the interval "does not match", the program offers its value in the message:

Suggested sample time (sec) = X.X is the recommended interval in seconds for the PID controller and the data collection for ExperTune.

The program offers a time interval with a margin - about 10 times less than the time lag of the process, that is, near the lowest, the "fastest" range limit. If we multiply this time by 2, we get a value close to the "slow" limit of the range.

Process Dead Time - process lag time

Process lag (dead time, clean delay) "Process Dead time" is the time required by the adjustable PV variable to begin to change after the valve position has changed. The regulator can not make the PV react before the process delay time ends.

Equivalent Dead Time - equivalent process lag

On the regulator side, it may seem that the process lag time is longer than its actual value. That is, the controller can not be tuned rigidly enough (but not leading to instability), so that the PV variable begins to react noticeably before the net process delay ends. More precisely, it can be said that the loop response time constant is determined by some "equivalent" delay of the process, which consists of a net delay plus an additional "tightening", which is introduced by the process components giving a phase lag of more than 180 degrees.

The delay phase increases in proportion to the frequency. Any process with a phase lag of more than 180 degrees has an "equivalent" lag.

Relative Response Time (RRT) - Relative Response Time

The parameter RRT serves as a relative indicator of the speed of the control loop. The smaller the RRT time, the faster the loop and vice versa, the slow loop has a large RRT value.

The RRT depends on the speed of the loop response to the load or job change and can be changed by changing the Safety Factor or the Setpoint Response time (Lambda) time.

ExperTune defines RRT in the following way (see figure):

 1) The frequency response of the closed loop response to the load jump is calculated. 2) There is a frequency at which the amplitude coefficient reaches a maximum

 3) This frequency is converted into a period of time.

Example of determining the RRT value of a circuit

On the amplitude-frequency characteristic there is a maximum (peak) at a frequency of 0.0677 Hz. Since 1 / 0.068 = 15 sec, the RRT of the loop in this example is 15 sec.

Often when using "aggressive", fast settings (Load-Fastest or Load-Fast), the PV response to load disturbances has the form of damped oscillations relative to the SP setting (see the trends in the second figure). The period of these oscillations is, roughly speaking, the same as RRT

The figures show that in the frequency response curve, as in the trend, the RRT time is 15 sec.


In this circuit, the fastest version of the Load-Fastest PI setting with a stability factor SF = 1 was used. (The process has a lag time of 6 sec and a clean dead time = 2). With these settings, the trend of the PV response to the stepwise load jump is in the form of damped oscillations. The time between the peak oscillations is approximately equal to 15 sec, which corresponds to the value of RRT.


Interactive loops To eliminate unnecessary loop interaction, set the stability level SF (for "Load" settings) or the response time constant SP Response / Lambda time (in the settings "To run the job") so that their RRT is 3 times different. Let, for example, there are 3 contours affecting one another, and the fastest of them has a time RRT = 1sec. Then reconfigure the other two circuits so that they get an RRT of at least 3 and 9 seconds.

Cascade control In the cascade control scheme, the primary (external) circuit should have a parameter RRT 3 times larger, that is, the circuit must be 3 times slower than the secondary (internal) circuit. First you need to configure a "fast" secondary circuit, then configure the primary circuit (master). If its RRT exceeds the RRT of the secondary circuit by less than 3 times, reconfigure the primary circuit.

How to ensure a good loop response at the same time to both load disturbance and SP job change?

To do this, you must install the SP target filter (setpoint) calculated by ExperTune in the loop. After this, SP can not be changed abruptly.

For many PID loops, a "Load" setting for better compensation of process disturbances (loads) results in a overshoot phenomenon when the regulator changes the SP reference. The filter smoothes the SP signal at the regulator input to relax the overshoot.

First, configure the loop for optimum Load Tuning compensation, and then use ExperTune to install the SP filter to eliminate overshoot when the job changes. For more information about the SP filter, see the corresponding section of the manual below.

PV filter in the control loop

The "Filter" option allows you to evaluate the need to install a variable PV filter in the loop, its type and characteristics for a particular process.

ExperTune calculates the optimal PV filter for each PID option of your regulator and enables you to see and analyze on the model the effect of this filter on the loop operation when changing the reference or load and the effect on the noise level. This is also useful for analyzing valve wear.

If the adjustable PV variable has a significant level of interference, you can learn on the model how to suppress, "smooth" this noise using one of the filter types provided by ExperTune. But do not try to use a filter to improve the quality of the data, as this will lead to incorrect calculation of PID settings.

The time constant of filter F should be chosen large enough to suppress noise as effectively as possible, without reducing the performance / quality of loop control. A too large value of F can disrupt the loop and reduce performance, since the control system will attempt to "adjust" the filter. Typically, the smallest user-level filtering level is the best option for the outline.

 Since the settings calculated by ExperTune are determined by the initial data presented in the Time Data for ... window, the user can always check how the PV filter affects the controller settings.

To set the PV filter in the contour model, select Trends from the menu: Edit- Filter ...

ATTENTION: Filtering changes the initial data of your process. When using the filter, all subsequent calculations of the controller settings will be based on the filtered PV data.

The last operation with the filter can be canceled using the Edit-Undo menu.

After selecting the Edit-Filter option under the Trend window, the PV Filter panel appears, where you need to enter the time constant F that you want to check and the type of filter.

Filter Time constant: The time in seconds that determines the properties of the adjustable variable (PV) filter, its transient response

Filter type: Selectable in the list shown.

Test: To test the effect of the selected filter on the path, click Test. After that, the trend of the PV in the Trend window will be recalculated and presented with the filter in mind. You can continue to enter other values ​​of F and, by clicking Test, look at the result of the filtering.

An analysis of the effect of the PV filter on contour operation.

IMPORTANT: The process input data used by the XTune program for contour analysis and calculation of settings must necessarily be collected without a PV filter. The filter can be removed automatically using the special function Un-filter PV (Remove filter).

In the window of the faceplate or in the window of the archive files (Off-line), after the command "Calculate the settings" (Tune), the "Find PV Filter" function becomes available. Select this function by placing a checkmark in the corresponding box to enable ExperTune to find the largest but acceptable filter for the selected setting. The filter should, on the one hand, suppress noise (noise) as much as possible, but at the same time it should not significantly reduce the efficiency of the loop's control (performance).

Select the type of filter from the list: 1st order filter, 2nd order, Butterworth or averaging filter. If necessary, you can select an engineering unit for the filter. This can be done from the Edit Setup window: the Advanced key - Loop Setup ... In fact, the PV time constant is the 4th PID adjustment parameter (F), which is represented on the HT faceplate with three other parameters of regulator adjustment: P, I, D. It is possible to talk about PIDF tuning of the control loop.

Try changing the current and new values ​​of the filter time and observe how thanks to the mathematical model the process response trends and the contour stability graph will be instantly updated depending on the type and size of the filter. Analyze the effect of the PV filter on the wear of the control valve using the noise analysis and modeling capabilities in the Control Loop Simulation window (see example below). In the "PID grid" settings window, review the entire range of PV filters for all settings, differently oriented for load tuning or setting change (Setpoint tuning), etc.


Types of filters

ExperTune allows you to use several types of filters.

The 1st order filter This is the most common type of filter. It is described by the following equation:

PVf = PV + LagTime * (PVf1 - PV) / (LagTime + SampleInterval), where

PV = adjustable variable LagTime = filter time constant found by the program ExperTune SampleInterval = interval (period) of reading PV values ​​and PVf calculation PVf = PV after filter PVf1 = last value PVf The units of the LagTime and SampleInterval parameters must be the same.

Second order filter This filter uses a cascade of two filters of the 1st order, with the time constant of each of them being equal to 0.5 of the set time F.

Butterworth 2nd order filter This is a very effective second-order filter. It is well suited for noise suppression, since it causes the smallest phase shift with better noise reduction compared to any other filter.

Averaging filter This filter acts on the principle of "moving average" - with the periodicity of the filter calculation, the average PV value is calculated for the previous time interval equal to the filter averaging time F. The effect of this parameter on the filtration efficiency is approximately 2 times less than the filter time constant 1- th order. Therefore, in order to obtain an averaging filter equivalent to a first-order filter, it is necessary to set the averaging time value to be twice as large as the filter time constant of the first order. If it is necessary to remove from the noise a certain frequency (harmonic), set the averaging time equal to the period of this harmonic and this should completely suppress it. In general, this type of filter at high frequencies is less efficient than Butterworth.

The best filter is Butterworth 2nd order, most efficiently suppressing high-frequency noise, although it is not so good at low frequencies. Behind him goes another good filter, the first order.

A properly designed filter reduces maintenance costs and extends the life of the control valve. If the filter is too "large" (large F) and significantly affects the low frequencies, this can worsen regulation.

Frequency characteristics of filters


The figure shows the amplitude-frequency and phase-frequency characteristics of the filters. The time values ​​of the filters are chosen so that all filters have the same value of the effective time constant, that is, the same phase shift at a low frequency. The characteristics of different filters are shown in different colors:

red - 1st order (In), time constant F = 10; yellow - 2nd order, two lag links with F = 5;

green-Butterworth (Btw), F = 10; blue-averaging with a window length of 20.

The ideal filter suppresses noise (noise) and does not cause a phase shift. The more noise is reduced, the better. The amplitude factor on the frequency response shows how the noise at a given frequency decreases. The smaller this factor, the greater the noise suppression.

Transient characteristics of filters


Together with the noise reduction, the filter causes a phase shift, which is bad, since it distorts the signal. The smaller the phase shift, the better. The effect of such a shift after installing the filter can be seen in the Trend window.

The requirements for the filter time constant F (PID-F)

The time constant of the filter should be:

• Less time of pure process delay (Process Dead time)

• Less than 0.1 time constant of the controller D 

Unique write location - A separate address for writing. If this flag is not checked, then the controller will use one common DDE address for reading and writing, if the flag is marked, then two DDE addresses. According to the DDE protocol, the full address must have the Application | Topic | Item structure

 Note. The APACS + / QUADLOG controllers have a function block FILTER, which performs the function of the 1st order filter and which can be used to set the PV filter. The block has a parameter TLAG - filter time constant in minutes.

The DDE server can be either an APACS I / O server (Application | Topic = APACS | DATASERVICES) or a Vision Framework application (Application | Topic = VIEW | TAGNAME).

The option of setting the time constant of the digital filter is also provided in the configuration parameters of the analog input channels of the SAI, SAM modules of the APACS + system.

SP Job Filter

The SP filter makes it possible to obtain a good transient response-the loop response

as with load perturbation (PV), and when changing the task SP. ExperTune finds the optimal filter for SP and, using an extensive database of different types of controllers, simulates and simulates the loop, allowing the user to quickly obtain and compare the transient characteristics and results of applying the SP filter.

The "Setpoint Filter" option is available in the Control Loop Simulation window, which is called after pressing the Analysis key.

Simulation of the SP filter and transient characteristics - loop responses.

  - show the response of the contour model when the job changes (Ortion | Setpoint Plot)

  - show the response of the contour model under load disturbance (Ortion | Load Upset)

  - show on the model the response of the loop to the change of the job with the SP filter and the new PID settings. This is equivalent to selecting: Setpoint Plot, then Option Setpoint Filter Use Lead / Lag Setpoint filter (Options | Filter SP | Use Lead / Lag filter)

   - Set the ratio of the time constants Lead and Lag (Option | Setpoint Filter | Adjust Lead / Lag Setpoint Filter = Options | Filter SP | Configure Lead / Lag filter). Lag time is automatically set by the program approximately equal to the integration time in seconds

After selecting the Use Setpoint Filter option, the program displays additional trends in the Contour simulation window for the outline with the SP filter: black color is the SP job, purple color- PV and CO.

Programming the SP filter.

Below are the equations for programming the lead / lag block that implements the SP filter function in the controller / controller.

Calculation Lag: y = SPuserSet + LagTime * (y1 - SPuserSet) / (LagTime + SampleInterval) Calculation Lead: SPcontroller = y + LeadTime / SampleInterval * (y - y1) y1 = y / * save for the next calculation cycle. At initialization, set y1 equal to the current value of the SP regulator.


SpuserSet = job entered by the user LagTime, LeadTime - time constants SampleInterval = lead lag calculation period y1 = last value y y = intermediate value (lag output) Spcontroller = value sent to the SP input of the controller

To implement the SP filter, use your company's lead / lag function block (for example, in the APACS + / QUADLOG system) or another programming option for these equations.

Summary table - Loop Summary Table - Loop Summary Table

A summary table (log) should help the user to select the best option for configuring the outline. The log is called from the faceplate or the Trend window with the key or from the options menu Options.

While working with the ExperTune program for testing and analysis, a certain influence is exerted on the circuit, various perturbations are introduced and appear. To analyze each perturbation, the Zoom In option is applied to the collected data, allowing you to select and process the desired part of the trend. As a result, depending on the data quality, linearity or non-linearity of the circuit, and possibly also on the direction (sign) of the controller's operation, the program gives out several different settings. Use the Log to register and view different values ​​of the settings of your circuit.


The settings log automatically calculates the average values ​​of all the settings parameters registered in the log, and also determines the most conservative variant of the contour setting. Some contours are non-linear and behave differently at different ends of the range. The others are asymmetric, and their reaction depends on the direction of the action or disturbance. For example, in temperature control loops, heating usually occurs faster than cooling. In such cases, you should analyze the operation of the loop at different ends of the range and / or in both directions and select the most conservative of the received settings.
Columns of the Pivot Table.

Archive: the number of the archive file of the circuit, according to which the values ​​of the settings are calculated

Zoom start, Zoom end: The data portion of the archive file (the trend section) used to get the settings.

P, I, D, F: Recommended values ​​of tuning factors (settings).

Fit: The quality of the raw data, determined by the frequency response

Gain: The process gain (transfer ratio) obtained by ExperTune

DT: Dead time (dead time, net delay) of the Dead time process found by the program.

Lag1: Lag- in the analog technology, the link is a smooth "tightening", the lag of the analog signal. Another name is the capacitive link. Lag time is the lag time constant of the link. If ExperTune has defined a process model with one or two lag links, then this column of the table indicates the Lag time of one or the first of the two lag links. Dynamic characteristics of the lag and lead links are shown in the figure below.

Lag2 or Lead: Lead - in the analog technology the link with the "advance" of the input analog signal (see the figure). If ExperTune has defined a process model with two lag links or with the lead link of the "advanced" process, here is the Lag time constant of the second lag link or the same parameter Lead time. The last (Lead) can only be in the case of a process with a reverse (inverse) reaction.

Intg: If the ExperTune model defines the process as "integrating", then this column puts the word "yes"

Lag time is the time constant of a capacitive link or a first order process. Time after a pure delay Dead time, during which the PV reaches a value of 63. 3% of its new stable state after a step change in the position of the valve. There are very few real processes that have only a lag lag, almost all of them contain a "pure delay" Dead time.

Dynamic characteristics of the links of the Lag tightening and Lead Lead of the analog signal

Stability: Relative stability index of the contour. The most conservative, that is, the most stable configuration option is assumed to be 100. All other settings are evaluated relative to this option. Their index is less than 100, it shows how much this setting is less conservative. For example, the index 50 shows that the setting is 2 times more aggressive than the most conservative vapriant. When you add a new setting to the log, the entire column can change.

RRT: Relative circuit response time for these settings.

Notes: Comments on each option. If the comment does not fit in the line, then by placing the cursor on this field, you can see the full comment.

Line of time units: The second line of the table header indicates the time units used in the ExperTune models. If there are different units in different models, the models are transformed and reduced to one, smaller unit. For example, if one model is represented in minutes and the other in seconds, then the first one will be converted into seconds.


Add - Add. A new line of PID and F parameter values ​​is added to the log, which are currently recorded in the column of new settings of the New control on the faceplate.

Remove - Delete the selected row.

Copy to New - Copy the values ​​of the PID and F parameters of the selected setting to the New column of the new controller settings.

View Archive - view the archive file that corresponds to the selected setting. The program opens the Trend of the data of this archive file. The same result can be obtained after double or right clicking on the first three columns of the Pivot Table.

 Report - Add the Pivot Table to the MS Word document.

Change Notes - Change the comment for the selected setting. The same can be done after double-clicking on the text of the comment.

In general, a double or right-click in the PivotTable window allows you to perform the following functions:

• Copy the values of the parameters P, I, D and F to the column New

• View archive file data

• Change comments

• Add the Table to the MS Word report

• Copy the window to the clipboard as an image file. Bmp, from where it can be inserted into any document

• Double clicking in the area of the first three columns automatically calls up the data view of the selected archive file.


This type of analysis is used to examine the normal operation of the circuit in a stable state with the SP setpoint unchanged.

Note. For a number of analysis tasks, such as statistical and frequency analysis of the contour, histogram and others, information is required on the stable operation of the loop without jumps and transient processes. Data collection for this case is different from other tests. The necessary data can be obtained in the following way:

• If the contour is stable, collect the data anew for analysis, using a semi-automatic procedure with manual data collection on and off.

• Using the Zoom In option, edit the Trend File information of the archive file of one of the previous tests of this circuit, selecting the necessary piece of data for stable contour analysis for analysis.

Statistical analysis and histogram

Collect data during the normal stable operation of the closed loop (without jumps and transients) before and after changing the settings.

Using the "Statistical Analysis" option in the Trend window, open the statistics window and see how the quality of the regulation improves with the new contour settings. Compare the statistical parameters "before" and "after".


1. Все статистические расчеты основаны на данных, представленных в Окне Трендов, а эти данные могут быть отредактированы, в том числе с применением функции Zoom in. Поэтому надо следить за тем, чтобы данные всегда были сопоставимы.
2. Объем собранных данных должен быть достаточным  для  достоверности  результатов анализа.
   Результаты статистического анализа контура представляются в табличной форме в специальном окне «Statistical Analysis», которое вызывается из Окна трендов. В таблице приведена статистика PV: среднее значение(Mean), стандартное отклонение (Standard deviation),  показатели изменчивости переменной PV(variability) и др, а также  статистика клапана (CO Statistics): показатели хода (пройденного пути) и реверса клапана



Normalized option: Selecting this option will normalize all statistical information, converting it in accordance with the percent scale 0 - 100%. This helps to compare the statistics of different contours, because all the main parameters: PVmin, PVmax, range, average, standard deviation, variance, IAE and valve travel will be presented in one scale.

Statistical parameters of the circuit

Sample (raw): The data collection interval for the selected archive file.

Num of points: The number of PV points collected

Range: The range of PV data values

Mean: the arithmetic mean of the collected data

Variance: Dispersion is a quantity that characterizes the spread of data values.

 Sample variance = sum (Mean-x (i)) ^ 2 / (npts-1

Standard deviation: The standard deviation of a random variable PV is the standard deviation from the mean, equal to the square root of the variance.

Variability: The variability of a random variable PV is the relative value of variance. Expressed as a percentage of the average and allows you to compare the level of variability of different processes.

 Variability = 2 * (Standard Deviation) * 100 / Mean

Variability Index: The variability index is a statistical measure of the efficiency (performance) of the regulator compared to the virtual "best" efficiency achieved with an "ideal" contour adjustment with the minimum possible variance. This index is calculated by the program and takes values ​​in the range from 0% to 100%, where 0% corresponds to the "ideal" with a minimum variance, and 100% is the worst case.

We can say that "variability" is a measure of "non-ideal" regulation.

IAE (Integrated Absolute Error): The real accumulated (total) absolute controller error is the PV-SP difference integral for the data presented in the Trends window. ExperTune calculates IAE only if the window contains data about the SP job of this loop. For example, an additional contour may not have a trend SP. The IAE parameter is a measure of the efficiency of the circuit. The IAE is equal to the area of ​​the Trends window part enclosed between the SP job curves and the PV variable. The smaller the IAE, the better, since this means that the contour is working closer to the task. The IAE parameter is useful in that it is easier to relate it to financial and economic indicators in different industrial enterprises than any other indicator of productivity.

In many cases, the control loops should work as close as possible to the requirements of the technical regulation (specification), with a minimum of deviations. Let, for example, in gasoline it is necessary to add an expensive additive MTBE so that its concentration is 2%. On the one hand, adding more than 2% of MTBE is a financial loss, but at the same time it is necessary to add enough to provide the required 2%. A typical approach: setting the SP is slightly higher than the 2% MTBE. Then, the better the circuit is tuned and the smaller the IAE, the closer to 2% you can set the loop and the less the MTBE flow.

To the right of the IAE parameter is a combo box with a list where you can select the time period for IAE extrapolation for an hour, day, week, month or year based on the data in the Trend window.

An example of a statistical analysis of the operation of a circuit under normal conditions with a constant value of the SP

Before configuring ExperTune

The PV trend shows the presence of cycles and load disturbances
After setting up

CO Statistics - Regulator and valve output statistics.

Travel: Valve stroke, stroke length,

Reversals: Number of valve reversals

The CO signal first passes through the pneumatic transducer, then the control mechanism moves the valve stem. ExperTune determines the movement of the rod by changing the CO and calculates the stroke length and the valve reversal. The valve travel is the estimation of the total distance traveled by the valve stem as a result of changes ("movements") of the CO signal presented on the graph in the Trend window. Reverses are the number of changes in the direction of movement of the rod, also occurring as a result of the change in CO over time.

Use the valve travel and reversal values ​​and contour modeling capabilities to analyze the contour and reduce valve wear. Data collection for these tests should be performed in the automatic mode of the regulator with normal, stable contour operation.

To the right of the "CO Statistics" table are the values ​​of the stroke length and the number of reversals for the selected period of time: hour, day, week, month, year. For example, if the valve stroke is 100% per day, this means that the valve passes a distance equal to its full range in 1 day. At the same time, it can, for example, move 1 time from one extreme position of 0% to the opposite 100%, or move 10 times back and forth between the positions of 50% and 60%.                        

Bar chart

The histogram is also constructed according to the results of statistical analysis and is called from the Trend window. It is possible that before that you will need to use the Zoom In option to select a section of the trend with the necessary "stable" data.

In general, for the construction of a Histogram, it is necessary to collect data on the normal operation of the circuit in a stable state. In the case of a closed loop (automatic controller mode), the SP reference must be fixed, in the manual mode, the output of the CO controller must be unchanged. In order to compare two histograms of one circuit, for example, before and after changing the settings, or in Manual and Auto modes, it is always necessary to use data at equal intervals

The histogram allows you to see and evaluate the statistical spread of your contour data. If the noise is "white", then the histogram will have a symmetrical bell shape. Place the cursor on one of the columns that make up the histogram, or above it, and in the small window the value% of the data corresponding to this column will appear.

Histogram of PV

 Collect the data in the automatic mode of the controller and call up the Histogram.

The shape of the approximating curve enveloping the histogram along the tops of the columns allows one to judge the quality of regulation and possible contour problems:

64.png ·         

- curve in the form of a high and narrow bell indicates a good adjustment of the circuit and a small deviation.

- a low and flat bell indicates a poorly tuned contour and a large deviation.


- if the histogram looks like an inverted bell, that is, with a hollow in the middle, this indicates a possible sticking of the valve.

The histogram on the left is obtained from data collected under normal stable contour conditions. The two humps on the left and on the right indicate that the PV values ​​are usually on one side of the SP, so you need to perform an adhesion test.


  Compare the histogram curves before and after adjustment. If the adjustment improves the regulation, the "bell" of the histogram should become higher.

  A similar property is valid for the Automatic and Manual modes of the contour. If the adjustment improves the adjustment, the histogram curve in the Auto mode should be higher. Circuits with cyclic oscillations and, especially, with the adhesion of the valve, often give in the histogram two "hump" on the left and right at the ends of the curve.

The histogram of the controller error (E = PV-SP)

Collect the data in the automatic mode of the loop. If the data in the histogram are grouped on one side of zero, this indicates that there is some kind of nonlinearity in the circuit.

The plot of PV versus CO (XY Plot)

The window with the graph of the PV variable (Y-axis) versus the output of the CO controller (X axis) is called up from the Data Trend window with the corresponding key or through the Options menu: Options - XY Plot. The scales of PV and CO are the same as in the Trend window. When you change the scale in the Trend window, the XY graph scale changes similarly. Each point (data pair) of data in the Trend window corresponds to one XY point of the graph. These points can be connected by lines (option in the Options menu).

XY graph can be useful for detecting oscillatory circuits. In the case of such a contour, there will be very few points in the center of the XY graph. The variant of the graphic image (points, lines, thick line) is selected in the Options menu. (see picture).


Auto- and Cross-correlation


Decreasing the area under the autocorrelation curve indicates that changing settings and other actions have improved regulation, making it less deterministic, more random.

In addition, autocorrelation helps detect cyclic fluctuations in the circuit - on the correlation graph, this is seen better than on the trend. For example, a 10-second cycle in the flow control loop is probably due to an unsuccessful regulator setting. If the cycle has a frequency of 2 or 3 Hz, then the possible cause is pulsation of the pump.

The confidence limits (2 / sqrt (Npts)) are shown on the graph in blue lines. Here Npts is the number of points, sqrt is the square root.

Cross-correlation of signals of two PVs

Correlation analysis helps to see and appreciate the degree of influence of one contour on another.

To analyze the cross-correlation of two PVs, select (mark) one of the PV variables as the output of the CO controller in the Setup window and in the Setup window.

Use cross-correlation to identify the presence of interaction, the relationship between the two circuits. If there is no dependence, then the values ​​of the cross-correlation graph will be close to zero.


The loop testing for spectral analysis differs from the tests designed to determine the controller settings. The collected data should not contain any special changes-the SP jump or the CO output signal, similar to the contour tuning tests. Data collection must be performed in a stable contour environment.

The spectral power density of the signal

Diagrams (graphs) of the spectral power density of the signal are called from the Trends window by the key or through the options menu: Options> Power. Then you can choose a variant of the spectrum:

• Deviation - the error spectrum of the controller E = PV-SP (not always available)

• PV - spectrum of PV.

• CO - spectrum of CO.

Spectral diagrams can be used to evaluate the improvement in loop performance as a result of changing the controller settings. The diagram shows the distribution of the relative power of the signal at different frequencies (harmonics) in the period range from a double interval of reading the data (sample interval) to twice the total test time.

  . They can be used to check the efficiency of the controller without having to test the circuit by changing SP or CO. The open-loop data power spectrum determines the point of discontinuity (transition) between frequencies carrying load disturbances and frequencies containing only minor noise. The goal here is to react only to disturbances.

One of the effective applications of spectral diagrams is the identification of any cyclic perturbations and their frequency components. Collect the PV variable data in the open loop (regulator in manual mode). The presence of peaks in the spectral density graph below the transition frequency should be considered suspicious. Is the cause of this peak a cyclical disturbance in the circuit? Or is the peak caused by mechanical vibration? Check the manifestations of this frequency in the process and find a cause-and-effect relationship with a similar frequency somewhere higher in the production flow chart ..

Another application of Spectral Diagrams is the evaluation of the quality of the regulator setting. Collect the PV data in a closed loop (automatic regulator mode) before and after adjustment. The smaller the amplitude of the values ​​on the diagram, the better the setting.

Option Spectrum of cumulative (cumulative) power Cumulative Power Spectrum. This type of frequency spectrum shows in percent the total power of all harmonics below this current frequency. The usual frequency spectrum does not always allow you to quickly find the contribution of a particular harmonic to the total noise power. Harmonics can give 10% or 90% of the noise power - the power spectrum alone does not show it. The cumulative spectrum shows that a certain percentage of cyclical fluctuations and perturbations are below this frequency.

Spectral frequency analysis is performed by applying the Fourier transform to the trend - the graph of the PV (t) or CO (t) function, which was obtained as a result of the acquisition of contour data.

Groups of peaks on the Spectrum diagram (Clustered Peaks)

Neighboring peaks, grouped around a large peak or several such peaks, form a "cluster" (group). ExperTune calculates the power of the cluster, summing the peaks of this group.

The spectral diagram shows the parameters of four such clusters. For each cluster, the following coordinates are indicated: coordinate along the X-axis in Hz (or period per second), as well as the percentage of occupied signal power in%.

Cluster power = the sum of the peak powers in a group. Cluster Coordinate = the mean value of the coordinates of all the peaks in the cluster weighted by the power of each peak.

Therefore, it is likely that the coordinate of the cluster will not coincide with the location of any one peak. If there is a large peak in the cluster, then the X coordinate (frequency) of the cluster will be close to the frequency of this peak. Using clusters allows you to more accurately determine the peaks and their power.

The Peaks in Zoom Area Only option.

Selecting this option causes XTune to use only those clusters of peaks that are in the selected (Zoom In) area of ​​the Diagram. By default, the search for the largest clusters of peaks is conducted across the entire spectrum.



Process and contour simulation, analysis of optimization results

Model and Analysis windows

After calling up the Tune control function (for the archive file) or at the end of the AutoTune auto setup on the faceplate, the Controller Tuning control panel opens, the Analysis button appears to the left of it, and the new recommended PIDF regulator settings are shown in the New column , that the quality of the source data is good enough to calculate the settings).

Clicking the Analysis key brings up four windows on the screen - four powerful contour analysis tools (see the figure below):

• Process Model Process Model Window

• Process Frequency Response process frequency window

• Control Loop Simulation window

• The Robustness Plot

When you change the selection of a process model, all four analysis windows are instantly updated to fit the new model. If you change any of the PID controller settings, the contour simulation window and the stability analysis window are updated, which depend on these parameters.

One of the most useful and effective analysis tools is the Control Loop Simulation control window, which shows the trend of the loop response with new settings before they are loaded into the regulator. More importantly, not only the response to the change in the SP task is modeled and shown, but also the response to the load upset disturbances. Usually, the user needs to check the loop response, eliminate or reduce exactly the random disturbances, load jumps occurring at the control object. In most cases, this is more important for the quality of the product and for reducing production losses than the optimal response to task changes.

The transient response (response) of the circuit when the task and the load varies is very different. On a working object, it is difficult to verify the operation of the loop when the load is disturbed, but this can easily be done with the help of ExperTune.

However, it must be borne in mind that any modeling is based on the use of a mathematical model of the process and directly depends on the quality of this model. Therefore, the choice of the process model is very important for the simulation results of the circuit and should be the first step in the analysis procedure. ExperTune provides two Laplace models, first and second order. The process model is based on the frequency characteristic of the process, which in turn is built on the basis of the collected test data presented in the Trend window. Thus, the quality of the models, the results of simulation and analysis of the contour, stability curves - all depends on the quality, reliability and compliance of the primary collected data.

In the Frequency characteristics window, the amplitude and phase characteristics of the real process and its model selected by the user are displayed simultaneously. By the degree of coincidence of the characteristics, it is clear how the chosen model corresponds to the real data. It is necessary to choose the model that best corresponds to the real frequency characteristics of the process, the most important frequency range, where the phase shift is from 90 to 180 degrees. When you change the model, all other analysis windows are instantly updated.

After selecting the process model in the Contour simulation window, the contour response of the contour model is displayed with the current and new PID controller settings. In practice, one always has to find a compromise between the speed (hard setting) and the stability of the contour to changes in the dynamics of the process. Here, the Stability Analysis window of the contour (Stability graph) helps. If there are two variants of PID settings with approximately the same degree of contour stability, then you can safely use a faster setting.

 Window of the mathematical model of the Process Model process

The window shows the process model found by ExperTune on the frequency response of the process and is represented in terms and notation of the Laplace transform and operational calculus in the form of an algebraic expression, where s is a complex variable.

 In the table under the formula of the model the values ​​of the dynamic parameters of the process are given:

• Gain - gain

• Dead time - delay

• Time constants - time constants

If the process is an integrator, then the time constant of the first order is designated as an "integrator" (see the figure). If the second-order model has imaginary roots, this is denoted by the word "imaginary".

Model Type Selection:

To determine the type of model, you must make a choice of the proposed options in two lists.

1. About using the coefficient Gain of a stable state.

The following choices are possible:

• Allow gain to float - Allow the Gain coefficient to change. The program ignores the information of the stable state of the process and builds a model corresponding to the frequency response at higher frequencies, which are more important for optimal and stable operation of the closed loop.

• Force steady state gain - Accept the steady state gain Gain. In this case, ExperTune takes the Gain equal to the value of the amplitude-frequency characteristic at the lowest frequency. In this case, the model is constructed by matching with the remaining "good" frequencies.

The value of the frequency response at the lowest (zero) frequency is not displayed at all, so it can happen that at the lowest frequency the characteristic of the model does not coincide with the characteristic of the process.

2. Order of the model

ExperTune automatically selects the use of a first or second order model, depending on the degree of coincidence of the frequency characteristics of the model and process. If the models differ insignificantly by this criterion, then the program uses only the first-order model. However, the user can force the order of the model. If, however, the data quality is not good enough, the message "Data not good enough to support this model structure" is issued instead of the model formula.

The following choices are possible:

• "Automatically Choose Best" - Choose the best option automatically. ExperTune chooses the best model itself, as described in the previous paragraph.

• "Force First Order With Dead Time" - Use a first-order model with the Dead Time gain values ​​found, gain, and the time constant Time constant

• "Force Second Order With Dead Time" - Use a second-order model (the formula contains the variable "s in square") with the found values ​​Dead time and Gain.

• "Force Integrator With Dead Time" - Use the integrator model with the found Dead time, Gain and integration time values.

If the option Inverse response process was selected in the Advanced window (called through the Edit Setup window: Advanced - Loop Setup ... - Advanced), the additional parameters Lead time and Integrator are displayed in the process model window (see example above and a description of the process of this type in the section "Loop problems"), and the order of the model is selected by ExperTune automatically - the option "Automatically Choose Best" is forced.

Option Start Simulator With This Model- The program loop simulator with the current model, PID algorithm and the current PID settings is started, provided that the program simulator ExperTune Loop Simulator is installed on the computer.

Frequency Process Response Window Process Frequency Response

If you feed a sinusoid to the input of a linear system (process), then the output of the PV process will also be a sinusoid, but, in general, with a different amplitude and, possibly, a time delay with respect to the input signal. The ratio of the amplitude of the signal at the output to the amplitude at the input is called the "dynamic gain" of the process or the "amplitude coefficient", and the time lag of the signal is called "phase shift" or simply "phase". The amplitude coefficient AK is measured in decibels, where dB = 20log (AK).

Different harmonics give different amplitude coefficients and phases, and the entire spectrum of harmonics allows us to construct the frequency characteristics of the process: amplitude frequency response and phase phase response.

The window shows the characteristics of the real process and its model (the selected type), which makes it easy to compare them.

Frequency characteristics are built on the basis of real loop testing data presented in the Trend window, so the quality of the model, like the PID settings, is highly dependent on the quality of the original data. In the general case, smooth curves that smoothly decrease from left to right with increasing frequency, correspond to good, qualitative data.

The better the collected raw data, the smoother the frequency response and the more accurate the model, the better the result will be calculated by the PID program settings


The actual process characteristic usually contains noise or peaks, irregularities at higher frequencies on the right.

The characteristics presented in the window help determine the best process model. For this, in general, it is necessary that the curve of the model correspond to the maximum, closer to the real characteristic of the process at frequencies where the phase shift lies in the range between approximately 90 and 180 degrees. The figure shows how to find these frequencies.


• Radians / sec or Cycles / sec - frequency unit: radians / sec or hertz. The default is a unit of radians / sec.

• Show High frequencies - show high frequencies. The program shows AFC and PFC for frequencies, where the phase is less than -270 degrees. Usually these high frequencies are not shown on the graph.

If f is the frequency in hertz, w is the angular frequency in rad / sec, then w = 2πf.

The angular frequency for the time constant T (sec) is defined as 1 / T radians / sec.

Control Loop Simulation.

In this window, the work of the circuit in a stable state and during transient processes is shown on the selected model-how the selected process model and controller react to external influences: job change, load jump, input noise, etc. with current and new PID settings and filters.

If the user modifies the process model or any parameter of the PID settings, the loop response trends are instantaneously recalculated, and new trend graphs are displayed in the window. This allows the user to try, compare and choose different options on the model.

In the upper half of the window, the trends of the adjustable PV variable and the SP setting are shown, in the lower half - the output of the regulator (see figure). The current and new settings use different color trends:

• Current settings: PV, CO - blue, SP - black

• New settings: PV, CO - red, SP-black

When using the SP: PV, CO filter (with SP filter) - purple, SP with filter - black

 Keys in the contour simulation window.

Note. All the main windows of ExperTune have a menu of Options options, which duplicate the function keys of this window. Therefore here are also given the English names of the keys (options).

  Load Plot- show the reaction to the shock / load disturbance. If the option "For Load" is selected in the Controller tuning window, this reaction is shown first.






With the help of the SP filter it is possible to provide a good response (transient characteristics) of the circuit simultaneously to changes in the target and to load jumps. For many circuits, adjusting the controller, which is optimal to compensate for load disturbances, causes an unacceptable overshoot when the job is changed. To attenuate, damp the overshoot, the SP filter smoothes the change in the SP signal before it enters the regulator input.

The Robustness Plot stability / stability analysis window

Contour stability graphs are an effective analysis tool. They graphically show the degree of stability (or sensitivity) of your circuit to changes in the dynamic parameters of the process - the Gain gain and the Dead Time lag, as well as the trade-off between speed and stability. This window is convenient for quick analysis of the stability / stability of the contour.

When tuning, you always have to find a compromise (balance) between speed and stability of the circuit, because a fast loop with a short reaction time is sensitive to changes in the dynamics of the process and is subject to fluctuations (the oscillatory circuit), and the stable contour, on the contrary, works slower.


The cross represents the point corresponding to the current process with the actual values ​​of the delay and the gain factor to which the controller is tuned. Changes in the process Gain gain and the proportionality factor of the controller P have the same effect on the stability of the closed loop. On the chart, you can see what happens to the contour stability, when the parameters of the DeadTime and Gain process are changed independently.

Comparative analysis shows two stability limits in this window: blue for current controller settings, and red for new settings (see figure below).

Typically, a stability margin that maintains the loop stability when the process parameter is changed by a factor of 2 (or SF times, where SF is the Factor of Safety Factor) relative to the current process (cross) is considered sufficient, "acceptable" for the circuit.

The points obtained by multiplying or dividing by 2 coordinates of the current process are represented on the graph by the apexes of the blue trapezoid, which shows the zone of permissible drift of the dynamics of the process. The vertices are connected by lines, which on a logarithmic scale become straight lines. In practice, in order to ensure the stability of the system, it is necessary to maintain the contour settings so that the stability boundary line always passes beyond the blue trapezium.

The stability graph of the circuit is calculated on the basis of the process model and regulator settings, the accuracy of the graph depends on the accuracy of the model. If you select a different model, the graph is updated.

If the contour is very stable or, on the contrary, very unstable, then the lines of stability boundaries can go beyond the limits of the graph.

Variants of the stability graph presentation In the Options menu, you can choose two options for representing the stability graph in simulation - in absolute or relative units of the process parameters Gain and Dead time:

Actual Gain and Dead Time - absolute (real) values ​​of the process parameters along the coordinate axes. This option is accepted by default.

Gain and Dead Time Ratio (relative values ​​of Gain and DeadTime) - representation of the graph in relative units of the coordinate axes, where

The relative value of the parameter Gain or DeadTime = (parameter value) / (the current value of the parameter to which the controller was tuned)

The cross on the graph, where both relative coefficients are equal to one, corresponds to the parameters of the process to which the contour was tuned.

Estimating the increase in contour stability with new settings

Above the graph in the special window the value of the parameter Robustness Increase is displayed - the expected calculated increase in stability in percent with new settings. If stability, on the contrary, decreases, then the value of Robustness Increase will be negative, and if the contour becomes unstable with the new settings, the designation N / A is given instead of the number. This parameter can be used to quickly, numerically enter the requirement in ExperTune how much stability should improve when changing to the new settings.

The calculation of Robustness Increase is based on comparing the values ​​of another parameter - Closest Distance with new and current settings (see below).

If you select the Gain and Dead Time Ratio option in the window above the graph, in addition to the Robustness Increase stability parameter, you can look at the parameter Closest Distance for the current (for the current) and new (for new) settings.

Closest Distance is the minimum distance, that is, the smallest gap in the stability graph (in relative units) between the cross of the current process and the stability boundary line. This parameter is a relative numerical measure of the stability of the contour, and quite conservative (cautious, taken with a reserve). If the stability boundary passes through a cross, this means that the contour is on the verge of instability.

When viewing the red values ​​of the parameter refer to the new settings, and blue values ​​to the current settings. If the contour is unstable, the value of Closest Distance = N / A.

Thus, the Robustness Increase and Closest Distance parameters give a numerical estimate stability, stability of the control loop.

How to enter the desired degree of stability in the system. With the mouse, the user can drag the red stability line to the left or right for new settings, thereby setting the desired degree of contour stability. In this case, the proportionality factor of the gain controller is automatically adjusted so as to obtain a new preset stability. Moving the stability boundary to the left reduces the stability margin, but gives a faster response (increases the gain). Moving to the right gives a more stable contour with a slower response. Since ExperTune also adjusts the integral I and differential D components of the PID settings, this adjustment for the Safety Factor factor for many circuits gives a better balance between stability and speed.

Processes with a low delay Dead time If the delay (dead time) of the Dead time process is close to the data collection interval for testing, then ExperTune estimates the amount of this delay very carefully, with a margin. Therefore, if your process has a very short delay time, its model will be less stable than the actual process. The line of stability boundary for the model will pass below, and the transition characteristics of the model in the analysis windows will also look less stable.

Examples of the influence of the regulator proportionality (gain) factor on the stability and operation of the circuit

1) The optimal balance between stability and reaction rate when compensating for load surges (load upset)

    2) The double proportionality factor of the controller P

3) The coefficient P of the regulator is increased by another 30%

Generalized performance indicators of the new recommended settings for the Performance Summary loop.
The Performance Summary window and the additional windows called up from this window with the "Evaluation" and "Time Line" buttons allow you to estimate the efficiency of the new recommended controller settings using simulation and generalized figures presented in numerical and graphical form compared to the current settings. To do this, the window presents generalized indicators that characterize the expected increase in productivity and improve (or, more accurately, change) other important parameters of the contour operation.

Adjustment of the regulator is always a compromise between various requirements: the performance and stability (stability) of the circuit, and, possibly, between the performance and the life of the valve. The window presents a generalized result of these tradeoffs implemented in the outline settings. To see all this in more detail and in a graphical form, click the "Evaluation" button.

Performance Increase - Increase the productivity, production efficiency of the circuit.

This indicator tells you how much better the PID controller will compensate for the load jumps at the new recommended settings. Usually it is proportional to the possible saving of money. For more details, see section 5 "Configuring the controller. Calculation of PID settings and filters "- Regulator adjustment panel (approx. Page 42).

Loop Performance - the effectiveness of the application of the control loop in the process, affecting the quality and cost of production. It is determined by the speed and quality of regulation in the circuit. Ultimately, this indicator is proportional to the economic efficiency of the circuit.

Robustness Increase - Increase stability (stability) of the contour

Estimation of the increase in the stability of the circuit, the expected calculated increase in stability in percent with the new settings. If stability decreases, the Robustness Increase value will be negative, and if the contour becomes unstable with the new settings, the designation N / A is given instead of the number.

For more details, see section 7 "Simulation and analysis of the contour" - Robustness Plot stability analysis window (approx. Pages 47-48)

Relative Response Time RRT - Relative Response Time of the loop.

Relative indicator of speed, response rate of the control loop to the disturbance. The smaller the RRT, the faster the performance and vice versa, the larger the RRT, the slower the circuit will work. This indicator is useful for different tasks of comparing contours, models and settings.

The value of RRT depends on the speed of the loop response: the user can change it by changing the safety factor or the Lambda time constant of the loop.

For more details, see section 5 "Configuring the controller. Calculation of PID settings and filters "- Control panel of the controller - Settings that provide the best compensation for changes in the Setpoint tuning task

Valve Travel Index - The index of reducing the stroke of the valve.

The percentage improvement (ie reduction) with the new settings for the total amount of movement of the control valve stem compared to the current settings. If, with new settings, including the PV filter, the valve stroke increases, the value of this index will be negative.

Valve Reversal Index- The index of reducing the number of valve reversals.

 The percentage improvement (ie reduction) with the new settings of the estimated number of changes in the control valve stem compared to the current settings. If, with new adjustment parameters, including the PV filter, the number of valve reversals increases, the value of this index will be negative.

Valve indices based on - Calculation of valve indices is performed according to the contour response trends in the Control Loop Simulation simulation window on the disturbance:

• Load-load button

• Setpoint - task change

• Noise - the noise in the PV signal

A window for evaluating the effectiveness of the recommended settings for Control Loop Performance Evaluation

1. The indicators of the valve dynamics are determined by the trend of the response of the contour model to the load jump


The window contains three sections with a graphical representation of the overall performance of the circuit in the form of linear indicators, columns of blue and red. The blue color refers to the current settings, and the red color corresponds to the new recommended contour settings. Selecting sections - in the options menu Options.

Section Performance and Robustness -Productivity and Stability.

The higher the columns in this section, the better. Columns - indicators of the performance of the contour are presented in percent relative to the "optimal" performance, taken as 100%. The red column (new settings) is inversely proportional to the current safety factor (safety factor):

                          Safety Factor             

1                                             100%                                     2                                             50%                                     2.5                                          40%
Usually, the user tries to optimize the outline so that all the red columns are as high as possible. However, it should be borne in mind that a 100% performance corresponds to the maximum speed of the circuit, and thus, its stability is likely to decrease.

The red column is not affected by the new values ​​that are entered manually, but the blue column is always set relatively red in accordance with the calculated Performance Increase performance. If this indicator is negative, then the blue column will be above red. In this case, under load disturbances, the current settings are likely to work better than new ones.

Columns-stability indicators of the contour are set equal to the value of the smallest gap Closest distance on the chart in the stability window of the contour Robustness Plot.

Section Valve Duty Diagnostics - Valve wear, valve performance, affecting its wear.

Valve Travel - Valve travel, the total distance that the valve stem passes during the transient response (the contour response trend in the simulation window). It is found by summing the absolute (without sign) value of the change in the position of the rod in each cycle of the test data collection. Comparison of the red and blue columns shows the effect of new settings on the intensity of movement and, consequently, wear rate and valve life.

Valve Reversals - The number of valve reversals, the number of changes in the direction of movement of the valve stem, corresponding to the trend of the loop response to the disturbance presented in the "Control Loop Simulation" simulation window. Usually, the more reversers, the faster the valve wears out.

Valve and contour improvement indicators

Valve Travel Index Valve Travel Index, Valve Reversal Index and Robustness Increase Robustness Increase estimate are generalized comparative percentages of loop performance improvement when changing from current settings to new settings.

For example, if the current stroke of the valve was 20, and with the new settings became equal to 10, then the valve life will double, that is, 100%. The valve travel index will be equal to 100%.

If, with the current settings, the valve stroke was 20, and with the new settings became 4, the valve life will be 5 times longer, and its stroke index will be 400%.

The considered numerical indicators of the intensity of movement - dynamics of the control valve, characterize the wear rate and, therefore, the service life of the valve. They can also be viewed in the contour simulation window in a special narrow display window to the right between CO and PV trends. Nearby are two small arrow keys, with which you can select the desired indicator from the list. The current values ​​are blue, and the values ​​for the new settings are red.

Comparing the current and new settings, you can estimate how quickly the valve will wear out. The smaller the stroke and the reverse, the less the valve wears out.

The output of the CO controller must first pass through the I / P air converter, then the valve mechanism moves the stem. ExperTune finds the stock movements, analyzing the CO signal, and then uses them to calculate the stroke and the number of valve reversals. Only such movements and reversals are taken into account, which change the position of the valve by an amount exceeding the dead zone of 0.1%.

Estimating the effectiveness of the recommended settings for Control Loop Performance Evaluation

2. Valve dynamics are determined by the reaction trend to the step change in the SP target


Valve wear can be reduced with a PV filter or by changing the PID settings.

First, if your system allows, try adding a PV filter to the circuit, using the filter time recommended by ExperTune. If this does not reduce valve wear sufficiently, consider modifying the regulator settings. In practice, there is always a trade-off between "fast" settings, loop stability and valve wear. To reduce the intensity of the valve movements as much as possible, remove the action of the differential component (parameter D) of the PID controller. After this, adjust the parameters P and I. Usually a decrease in the proportionality factor P gives a greater effect than a decrease in the integration time constant I.

 To get the minimum wear of the valve, select the option for setting the PI regulator with a large safety factor or a long reaction time. At the same time, the stability of the circuit will be good, and a compromise is achieved due to a worsening of the reaction to load disturbances and a change in the task.

The intensity of the valve movements The smaller the displacement and reverse movements, the better for the valve. To get the best analysis of the valve dynamics, set the Control Loop Simulation window to "Measurement Noise Response", and the valve values ​​will correspond to the actual conditions of the loop. Compare the valve wear values ​​for old and new settings. These numbers can also characterize the effectiveness of the PV filter.

Estimating the effectiveness of the recommended settings for Control Loop Performance Evaluation

3. Valve dynamics are determined by the trend of the loop model response to noise in the PV signal 

  Analysis of the relationship of time parameters in the control loop

The Control Loop Time Line Analysis window helps to understand and evaluate the timing relationships of all components that determine the dynamic properties of the control loop (see figure). For optimum and stable operation of the circuit, each of its time parameters must correspond to other dynamic characteristics of the circuit. For example, the controller / controller loop (calculation of the PID algorithm) should always be less than the deadtime (dead) delay time of the Deadtime loop. If the loop of the controller is close to Deadtime, then its reduction will significantly improve the circuit's performance. 


For optimal loop operation, the values ​​of the loop time characteristics must be located on the time axis in the following order:

 * Controller cycle * Filter time * Time constant of differentiation * Dead time of the process Dead Time * Equivalent dead time of the process Dead Time (in the case of the 2nd order process) * Integration time constant * Relative response time of the RRT loop

The value of each of these parameters, if it is not zero, is marked on the time axis in seconds in a logarithmic scale. Therefore, it is easy to compare the time of different components of the control loop. In this case, the integration time constants (I) and differentiation (D) are taken from the "New" column of the new recommended settings on the faceplate of this circuit.

 To facilitate comparison, the settings of the serial or parallel variant of the PID algorithm formula representation are given to one equivalent structure of the "ideal" (ISA) PID algorithm (if your regulator no longer uses the ISA option) and is represented in seconds.

In these units the following relation must always be satisfied: "The time of differentiation is not greater than the integration time," that is, D <= I. Otherwise, the differentiation (or filter) begins to interfere with the effect of integration. In some rare cases of contours with two large lag links, applying the value D> I can slightly improve the loop response to load disturbance, but due to stability. The ExperTune settings always give equivalent values ​​of D <I.

The formulas for calculating the three variants of the PID algorithm are as follows:


where Kc, Kp - coefficient of proportionality (gain), I, Ip - integral coefficient, D, Dp - differential coefficient of PID settings. AT

In the case of a second-order process (containing 2 lag units), an equivalent delay time is calculated and marked on the time axis using the formula:

Equivalent Dead Time = Dead Time + less of the two time constant lag links

To see the exact time value for a particular parameter, position the cursor at the desired location on the screen at the mark of this parameter.

Evaluation of time characteristics

As stated above, the loop time parameters must be located on the time axis in a certain sequence. Where possible, ExperTune monitors and evaluates the relative values ​​of time parameters and places colored strips with messages in the appropriate places in the diagram (see the figure). The color of the strip has the following meaning:

Green - everything is OK, OK Yellow - Attention, the contour can be improved here Red - The capacity of the circuit can be significantly increased.

Sometimes there is not enough room on the color bar for the full text of the evaluation message. In this case, place the cursor on the text to read the full message. You can also stretch the image horizontally and vertically.

ExperTune checks and evaluates the following timing relationships:

1) Controller cycle with respect to the Dead Time contour delay 2) The PV filter time is compared with the controller differentiation time constant 3) The differentiation time D with respect to the integration time I 4) The PV filter time is compared with the Dead Time delay or equivalent delay (for 2nd order processes0 All evaluations are performed only for nonzero parameter values


Special tests, diagnostics and recommendations


The control valve is the most typical and common source of hysteresis in the control loop, but not the only one. Hysteresis can be present in mechanical connections and the like.

The phenomenon of hysteresis is manifested when the inertial "sticking" valve does not manage to react to the change in the direction of the control pneumatic signal and continues to move for some time in the former direction, although the input signal has already changed to the opposite one. In this case, the valve will not start moving back until the input signal changes in a new direction by a certain amount in%, which is taken as the numerical value of the hysteresis.

Best practices from ExperTune

Hysteresis: If the system has a hysteresis greater than 1% for valves with positioners or more than 3% for valves without positioners, you should consider repairing or replacing equipment to reduce hysteresis and improve regulation. Often the installation of a valve positioner solves the problem.

Valve parameters: ExperTune believes that a well-designed workflow must have a process gain of 0.5 to 3. If this is not the case for your process, you may need to adjust: 1) valve size, 2) valve characteristics or 3) the sensor range.

Noise level: An interference level of less than 2% is considered "normal". However, the lower the PV noise, the better. If the noise exceeds 2%, consider using the PV signal filter to reduce interference.

To test for hysteresis, it is necessary in the manual mode of the controller to first collect a small part of the data on the normal noise in the loop, and then make several changes-jumps in the CO output signal: two steps in one direction and one step in the other.

Steps should be greater than 1% of the CO range. If the noise at the controller output exceeds 1%, the hysteresis test program will take it for the test step. To prevent this from happening, average the CO signal between the steps, clearing it of noise. 

Data collection should begin with a stable state of the regulated variable PV, which must then reach a stable state and after each subsequent step. Steps must be made quickly in a jump, since the CO output must accept a new value within a maximum of 3 data reading intervals.

The test procedure.

 1) Set the regulator to manual mode, wait for the stable state of the adjustable

 variable PV and enable data collection (archiving).

 2) Collect some data about the normal noise of the circuit, and then quickly increase the output

 of the CO regulator by 5%.

 3) Wait until the PV calms down and again quickly increase the CO output by 5%.

 4) Wait again until the PV calms down and quickly reduces the CO by 10%.

 5) Wait until the PV reaches a stable state and turn off the data collection.

  6) In the Data trend window using the Zoom in function, select the collected data of 3 steps and call the "Hysteresis check" function with the softkey or the Options menu.     


ExperTune determines the hysteresis using the following procedure:

• Finds and marks each of the three steps of the controller's CO (in Manual mode) with c1, c2, and c3, provided that the step-jump exceeds 1%. If there is a problem with some step, the program marks the last "good" step. If there is noise at the output of the CO controller, the data is averaged to clear the noise sections of the trend between the test steps.

• According to the initial test data collected with a stable contour state, the program determines the noise band of the loop, which is limited to the top and bottom by purple lines denoting the maximum and minimum of noise. Above or below, the inscription "noiseband" is given.

• Finds and selects a trend corresponding to a stable contour state after each of the three test steps. At the same time, through each of these sections, a straight purple line is drawn. If there is a problem with the "stability zone", the program marks the last "good" section.

• The red vertical lines labeled "p2" and "p3" are drawn between the three stability zones, as shown in the figure above.

• The hysteresis value in% is calculated and displayed. If you can not find the hysteresis from the collected data, an error message is displayed.

Option "Manually Choose Areas for Hysteresis Check" (Manual selection of data for hysteresis control)

This option can be used when the automatic test has problems with selecting suitable data areas in the CO or PV trend. To check for hysteresis, you must always manually make several changes to the output of the controller: 2 steps in one direction and 1 step in the other. Before the first step, it is necessary to collect some of the data on the normal process noise in the open loop (Manual mode of the regulator).

The option is selected in the Hysteresis Check menu of the Trend window and allows you to manually select the desired trend areas.

In the Trends window, there are 4 vertical lines, which you need to "drag" to "stable" sections of the trend using the mouse. The first line is placed in a stable zone close to the first change in CO, the next two lines are similarly placed before the 2nd and 3rd steps, and the last one is after the third step closer to the end of the trend.

The noise band is determined by the first stable data segment - from the beginning of the trend to the 1st line. ExperTune uses the CO and PV values at the points where the lines pass to perform a hysteresis test. Calculations of hysteresis, valve and sensor characteristics can be highly dependent on the location of these lines.

Is the valve size too high or is the sensor range too small?
In the flow control loop, the most likely reason why the process gain / gain can be greater than 3 is an oversized valve size. Such a valve can significantly impair regulation.
Any valve or actuator is characterized by the resolution or amplitude of changes in the adjustable value (range ability), which determine how accurate the adjustment will be. For example, a two-way valve has only two states and very low resolution. A good control valve can have an amplitude of 100: 1, therefore, the output of the regulator can be set with an accuracy of 1%. This means that the valve resolution is 1%. If the valve size is too high, then its useful working amplitude decreases. With a process transfer ratio of 2, a control valve with a potential amplitude of 100: 1 will have a real amplitude of 50: 1, which corresponds to a resolution of 2%.

If the size of the valve is too high, then it may happen that the valve is working in the limit position, on the “seat”. For the same reason, cyclic oscillations may occur in the circuit, since the control loop always fluctuates, “moves” around the equilibrium position within the valve resolution, or the valve can be on or around the “saddle”.
The problem of “large” valve size can be solved by fitting it or replacing the valve with a suitable size.
The process gain can be high even in the case of an incorrect valve characteristic. It is possible that the coefficient in this process zone is increased because the valve is linear, but it must be of the type “equal percentage” or vice versa.
Another reason for the increased loop transfer rate may be too small or a narrow nominal sensor range. Here to solve the problem it is necessary to increase the range. Note that this case is rare, since users usually try to have a larger range of sensors.

Is the sensor's nominal range too large or the valve size too low?
In the flow control loop, the most likely reason why the process gain / loop gain may be less than 0.5 is that the sensor range of the adjustable variable PV is too wide. .
For example, a 12-bit analog-to-digital converter (ADC) has a resolution of 1/4096 or 0.02%. If the sensor is used to measure temperatures in the range from 1 to 4096 degrees F, then measurement and regulation are possible with an accuracy not higher than 1 degrees F. However, if this sensor measures the temperature from 100 to 500 degrees F, then its resolution will increase 10 times and will be 0.1 degrees F. Thus, by reducing the nominal range of the PV sensor to match the actual operating range of the variable being measured, the accuracy and quality of the regulation can be improved.

A low valve size may affect the safety of the object, since its stroke range may not be sufficient to regulate the process. A possible solution to this problem is to “fit” the valve or replace it with a valve of the required size.
In addition, the loop transfer coefficient may be too small in a certain process zone due to an incorrect valve characteristic: a linear valve, but it must be of a different type - “equal percentage” or vice versa.

How to react to the error message: "Each CO change must be at its final value in 3 samples" (Each change of CO must occur quickly, within no more than 3 data collection intervals)
When testing for hysteresis, the program monitors the time of change of the signal at the regulator output. If the CO output changes too slowly or the signal goes with a high level of interference, the following options are recommended:
• Manually select data sections for hysteresis calculations in the Trends Window
• Repeat data collection.
• Edit the trend so that the CO output changes abruptly.
If there is an unacceptable level of noise at the output of the regulator, use the averaging function to clear the trend between steps.


The term "sticking" refers to the resistance to the beginning of the movement of the valve. Sticking or uneven, abrupt movement occurs due to too tight installation, improperly selected actuator or damaged valve stem corrosion. A very common defect is sticking on the “saddle”, especially for valves with tight closures.
Due to sticking, the control valve does not immediately begin to move under the influence of air pressure and eventually sets to the wrong position. Therefore, sticking always leads to cyclical fluctuations of the controlled technological variable.

Sticking is a big problem for the regulating circuit. The regulator causes the valve to move until the process variable PV reaches the SP reference, but due to adherence, the valve will continue moving and the variable will go too far, beyond the reference. Therefore, the regulator will change the direction of movement of the valve to the opposite and everything will repeat again. Such a limiting cycle, when the valve gets stuck and then suddenly slips when the input signal changes, is called an intermittent or abrupt cycle.
There is no general method to determine what degree of adhesion is permissible in a given contour. It depends on the particular circuit and the whole process. For many processes, sticking of 0.4% or 0.5% is too much. Adhesion ensures contour cyclical fluctuations and increased variability.

It is important to determine if there is adhesion in the contour and to what extent. After that, it will be possible to assess how much this sticking is harmful (or destructive) for the whole process. Sticking is the most harmful phenomenon of all possible valve problems. For example, hysteresis is, of course, an undesirable phenomenon, but usually it does not become a completely unsolvable problem. Another example is a non-linear valve characteristic, which can be compensated for by the linearization function installed in a regulator or valve positioner.

Usually, the degree of valve sticking during bench tests turns out to be greater than when working on an object in a control loop, as there is no lubrication, vibration, “noise” on the bench, and the force required for the valve to overcome the resistance of the process product, which also affects the operation valve and sticking. ExperTune program allows you to measure and evaluate the degree of valve sticking (in%), for “on the bench” and “loop operation” modes.

Data collection to check valve sticking

To check the contour for valve sticking, it is necessary to make several test changes in the CO output signal in the Manual mode of the regulator: one large step of 4% -5% and then several small steps from 0.1% to 0.3%. All steps must be in the same direction. Before the first step, it is necessary to collect some of the data on the stable operation of the open loop (manual mode of the regulator) without jumps and transients.

After each step, the adjustable variable PV should calm down and reach a stable state before the next step, with at least 30 seconds between the steps. Steps need to be done quickly, abruptly.
  The first big step of changing CO is to overcome any possible hysteresis in the circuit.
Note. The valve sticking test can be performed immediately after the hysteresis test. To do this, make a series of small steps - changes (0.5%) of CO output in the same direction as the last CO jump in the hysteresis test.
Data collection procedure for checking valve sticking

1. Place the controller in manual mode, wait for the process (variable PV)
calm down and turn on XT faceplate data collection and logging (the "write" key).
2. Collect the normal PV noise for 45 seconds in a stable state of the circuit, then quickly increase the output of the CO controller by 5%.
3. Wait for the PV variable to calm down (wait at least 30 seconds) and quickly increase the CO output by another 0.2%.
4. Repeat step 3 until the PV changes (the valve moves) after the next step (see example below), then wait at least 30 seconds and turn off data collection.
5. If necessary, in the Trend window, select (Zoom in function) the data of the performed steps and the initial section with the data of the stable contour operation.
6. In the Options menu of the trend window, select Stiction check. The program will analyze the collected data for the presence of valve sticking in the circuit.

Example of data collection for diagnosing valve sticking


Analysis of the collected data

Analysis of test data is performed in 2 stages and each has its own separate page - the image in the Trends window. First, on the 1st page it is necessary to mark with a vertical line the start time of each step — the change in the output signal of the CO controller. The program tries to do this automatically, but you may need to manually correct and clarify the location of these lines. If the CO signal has no noise, then it is likely that XTune will set the vertical lines correctly.
The user can move (Move), add (Add) or delete (Delete) lines with the mouse. To select a point (coordinates along the time axis) or a line, place the cursor on this object and double-click it with the left key or simply click with the right button.
When the installation of the vertical lines is completed, click the Next button. It is recommended to expand the Trends window in order to facilitate the work with the data.

Further, on the 2nd page, the program draws horizontal lines on the PV trend, corresponding to the average value of the PV variable in each trend segment between the steps - CO changes. As a result, each horizontal (stable) segment of the CO trend will have a corresponding PV trend segment, initially marked with a yellow and then a red horizontal line. (see picture below). The red line shows the average value of PV in this area, the yellow line extends the red line to the end point of the previous red line. This makes it easier to notice changes in PV with small changes in CO, which in turn are necessary to detect and measure small adhesions. When the valve shifts the sticking field, the PV variable may change by a very small amount. Indication of the mean PV value will help detect small changes (in response to a small CO step) against the background of the PV noise.

Snapping Test Data Analysis Example - Page 2


As can be seen in the figure, the program automatically placed 2 vertical lines on the trends in order to automatically mark CO values ​​corresponding to the valve sticking zone. If necessary, you need to more accurately set these lines manually. As a result, the first line should stand on the “stable” part of the trend before the 1st small CO step. XTune is likely to set this line correctly. The second line should mark a stable part of the trend after a real change in PV as a result of the next small step of CO. This second line marks the end of the sticking zone.
(see figure).
The difference between the value of the CO signal at this point and the value of CO before the first small step of the test determines the amount of sticking. XTune converts it to a percentage of the full range of the regulator output.
In fact, the actual amount of sticking is slightly less, since the last small step of the CO, which caused the PV to change, had to move the valve stem to overcome the sticking.

How to deal with sticking.

1. The best way to get rid of sticking is to repair the valve or the positioner. Ensure that the valve or positioner is supplied with a nominal air pressure for which it is designed. Some modern pneumatic actuators require a force of the order of 36.3 kg (80lb). Check that the size of the positioner matches the valve. If necessary, remove the valve / positioner and repair it.
2. Temporary or partial measures. Below are some of the techniques and methods that should be considered as partial and temporary. To really solve the problem, it is necessary to eliminate the very reason for sticking. These measures are always a compromise that will adversely affect the performance of the circuit or the service life of the valve. Methods that cause the valve to move more are likely to lead to additional valve wear, and a change in the optimal regulator settings will cause a decrease in loop performance. However, this may be ultimately better for the process than cycling due to sticking.

Adding noise to the regulator output Constant movement can help combat the intermittent, abrupt movement of the valve. In a system with a liquid material, add a periodic signal with a frequency of 300 Hz and some derivative in the PID controller settings. This will add noise to the signal at the output of the regulator. Reconfigure the positioner. Integration in the positioner in combination with sticking causes cyclic valve oscillations. If the positioner is a PID controller, use the derived component, but do not use the integrating action. In this case, the positioner will not be so accurate, but in it the cyclic oscillations will disappear. The derivative will help keep the positioner in motion, which may eliminate the intermittent nature of valve movement (while accelerating valve wear).

Using the PID regulator with the dead zone in the integral component The integral component of the regulator output signal in combination with valve sticking causes cyclic oscillations in the circuit. Use a PID regulator with a variable integral effect. Set the “dead zone” in the integral component so that with a sufficiently small PV-SP error, integration is turned off, that is, if | Error | <k, then Integral = 0. This approach is better than using the dead zone to the entire PID algorithm, as suggested below in the last paragraph.
Resetting the PID controller. The integral component of the regulator, in combination with the valve sticking, causes cyclic oscillations in the circuit. Remove the integral component, and the regulation will deteriorate, since a non-zero residual deviation of the PV from the SP reference will appear in the circuit at a constant SP (offset), but then the cyclic oscillations caused by integration will disappear.
Using the PID controller with dead band Set a small dead band for the PID algorithm. With a sufficiently small error, the regulator considers the error equal to zero, and the CO output does not change. In fact, introducing a dead zone regulator is a bad practice, but perhaps this will help overcome the sticking of the valve.


Many control loops are difficult to adjust because they are non-linear. This means that the process transfer ratio (Process Gain) changes depending on the PV value or the output of the CO controller. If you do not use linearization, you will need to adjust the regulator to the conditions when the transmission coefficient is maximum. As a result, the settings for the rest of the range will be “slow.” Linearization of such contours will improve regulation, since the regulator will be better tuned over the entire range.
If your process (variable PV) produces cyclical oscillations at one end of the range and turns out to be slow, inertial at the other end of the range, then all this is due to nonlinearity. Using the linearization block (ExperTune linearizer), you can achieve the same loop performance over the entire range of the adjustable variable PV, which will ensure optimal product production and eliminate cyclic oscillations (see the example below).

Nonlinear contour before linearization

Contour after installing the output linearizer


There are two types of linearizers at their place of installation in the control loop: "input" and "output" linearizers.
ExperTune provides the creation and use of both output and input (pH) linearizers.
Most circuits use an “output” linearizer that can be used with various non-linear control loops: flow rates, plating temperatures of chemical reactors, internal (secondary) cascade circuits, and generally in any loop where SP is a variable.
The output linearizer can be useful for almost all temperature, level and pressure control loops that directly control the valve.
  The “input” or pH linearizer is used to linearize the control loops of individual technological processes:
• indicator of the concentration of hydrogen ions (pH pH);
• temperatures of some distillation columns, in which the temperature profile is sharply bent in the center of the curve.
• redox processes

To find out which version of the linearizer your circuit needs, you need to check the reaction of the circuit (for example, to change the task) or its linearity. If the response varies with load or performance, it means that you need an output linearizer.
The guide then discusses the output linearizer.

 The output linearizer receives the signal from the output of the regulator and converts its value so that the contour remains linear over the entire control range. In this case, the circuit with the linearizer should ensure stability and optimality of settings on the whole scale of CO or PV. Testing the valve along with the rest of the process ensures that the XTune linearizer fits well with the valve and process characteristics. Testing only one valve without the rest of the process does not guarantee this, since too much depends on other elements of the control loop.
As a result of the analysis of the collected data, ExperTune creates and offers the user a linearization program for use in the linearizer. Then you can copy and paste this program into any Windows application (for example, ProcessSuite 4-mation).

Steps to create an output linearizer.
1. Data collection for linearization
2. Analysis of the characteristics of the process.
3 Selection of piecewise linear or hyperbolic linearization.
4. Select programming language
5. Setting up a linearized controller

Step 1. Collect and analyze data for linearization.
To control the nonlinearity, it is necessary to collect data on the stable operation of the circuit at several points in the range of the output signal of the regulator in Manual or Automatic mode.


An example of testing a nonlinear flow control loop is a nonlinear process response (PV) to changes in the output of the CO controller in manual mode.

• In manual mode, set the controller output to 5% of the CO range or
 in Automatic mode, set the SP reference to 5% of the PV scale
• Wait until the process calms down and reaches a stable state, when the trends of PV and CO are simultaneously straight horizontal lines (not counting noise)
• Increase CO output (or SP target) by 15%.
• Wait again for the circuit to calm down and reach a stable state.

Repeat this procedure in increments of 15% until you reach a value of 95%.
This will give 7 measurements of the “stable” CO and PV data at points 5%, 20%, 35%, 50%, 65%, 80% and 95% of the regulator range.
Comment. In automatic mode, make sure that the test includes the minimum (0%) and maximum (100%) acceptable values of the SP reference.


In the Options menu, select Characterizer. Then alternately select each stable part of the trend, marking it with a vertical line. Next, double-click or right-click and select Add.


Work with vertical markers.

Add line (Add). Place the cursor on the stable part of the trend that you want to mark, and right-click or double-click it. In the menu that appears, select Add.
The dashed line highlights the selected vertical line.
Move the line (Move). Place the cursor on the line, with the cursor image changing to a double horizontal east-west arrow. Click and hold the left mouse button and move the cursor along with the line.
Delete line (Delete). Place the cursor on the line, the cursor will change to
double horizontal east-west arrow. After double or right click, select Delete.

Step 2. Analysis of the process characteristics

The process characteristic is a plot of the adjustable variable PV as a function of the control output to the valve. If the quality of the collected data is good, the characteristic shows how non-linear the process is.

In this example, it can be seen that at different points in the range, identical changes in the valve control signal (1 and 2) cause different changes in PV: the process response in section 2 is much larger than the response in section 1.

The PV scale is shown in the figure to the left, while a part of the graph or the entire area corresponding to the PV range can be highlighted in gray. Areas above and below the PV range are highlighted in white. The scale of CO values is shown below (coordinate X).

In this step, the XTune program calculates and displays the minimum and maximum transfer gain (gain) values ​​of the Process Gain (PrG) process and their ratio.
 A smaller coefficient corresponds to a smaller slope, the maximum coefficient corresponds to the greatest slope, and their ratio characterizes the degree of nonlinearity of the contour.
The ratio of the maximum coefficient to the minimum should not exceed 3, and better not more than 2. If this ratio is greater than 3, then the output linearizer should be added (or modified) to the circuit of the circuit.
As a result of analyzing the characteristics of the process, XTune gives a message about the degree of nonlinearity of the contour and the recommended measures:


A split range control loop is a circuit that controls two or more valves, switching from one to the other, depending on the value of the regulator output. Usually, switching occurs at 50% of the CO output scale. For example, up to 50% of the circuit is cooled with cold water or heat exchanging oil, and above 50% it provides heating with steam, hot water or hot oil. Such contours, as a rule, are very non-linear, and the output linearizer is extremely useful for them.
Do not use the output linearizer with pH circuits - they need an “input” linearizer.

If the degree of nonlinearity of the contour is from 2 to 3 or more, the use of a linearizer will improve the work of the contour and will be effective.

Step 3. Build the output linearizer

After collecting the process data and marking 5-10 stable sections in the Trend window, you can use the Build button to call the linearization wizard, which will help to build a linearizer for your non-linear contour.
First you need to choose the variant of the linearizer, which is determined by the type of curve approximating the process characteristic:
• Piecewise Linear Fit - piecewise linear curve, which is defined by the X, Y coordinate pairs of the break points. This type is often used in industrial process control systems.
• Hyperbolic Fit — a hyperbolic linearizer that uses a single hyperbola formula, which allows you to build a linearizer based on just one division unit


Piecewise linear linearizer.
The characteristic of the process in the linearizer is described by a piecewise linear curve consisting of several segments. Initially, the user determines the number and coordinates of the linear segments, marking the breakpoints with a marker-red square, and the beginning and end of the entire curve (the linearizer input signal range) with a blue square. These markers can be easily added, deleted, or moved with the mouse after a “right” or double click.

Adding a marker. Point the cursor (mouse) to the location point of the new square marker, then double-click or right-click and select Add.
Delete the marker. Select the desired square with the cursor, the cursor type being named. Further
Double-click or right-click and select Delete.
Move the marker. Place the cursor on the desired square, with the view of the cursor being named. Then click the left mouse button and, holding it down, move the marker to a new location.
Automatic installation of markers. If you press the Avto Adjust key, XTune will try to place the red markers in the best way possible. Then it interpolates the linear segments between the markers and adjusts their vertical position. The user can shift the markers along the X axis and call the Avto Adjust function again to reset them along the Y axis.
To improve the linearizer after Avto Adjust, you need to manually adjust the markers again.

Hyperbolic linearizer
This linearizer uses only one equation, to solve which it suffices to perform several arithmetic operations, including just one division:

Valve =  CO /[L + (1-L)CO],     
where CO = regulator output, Valve = linearizer output, L = linearity parameter (constant).       

Try moving the red or blue marker with the mouse. At the same time, the hyperbola is changed and at the same time the screen shows the current value of the linearity parameter L for this curve. Adjust the curve (and the parameter L) so that the hyperbole maximally coincides with the characteristic of the process, that is, passes through all the crosses.


Setting the linearizer by changing the parameter L. One of the advantages of the hyperbolic linearizer is that you can “tune” it in place, directly on the control object by “fitting” the parameter L. At the same time, there is no need to follow the whole process characteristic - if the contour generates oscillations during low values ​​of CO, increase L, if oscillations occur in the upper part of the CO range, it is necessary to reduce L. Note: after changes in the parameter L, the gain of the regulator must be adjusted.

Adjusting the proportionality coefficient (gain) of the controller after changes in L.
As a rule, even before linearization, the regulator is already configured to stabilize the circuit with the CO output value, when the loop gain (closed circuit, including the process) is maximum. Therefore, if the linearizer is used correctly, it reduces the loop gain with this value of the CO output and allows to increase the proportionality factor of the regulator in the same proportion (in the APACS + / QUADLOG system it is referred to as PG). The gain of the linearizer at the 0% CO point is 1 / L, and at the 100% CO point it is L. Therefore, if the L <1 value is set to eliminate oscillations at 100% CO, then the proportionality factor of the regulator can be increased by multiplying by 1 / L. On the contrary, if it is set L> 1, then P can be increased L times.

Control systems with CALC computing blocks that work with variables and constants on a scale of 0% - 100% (0 - 1).
In this case, for the Hyperbolic linearizer, 2 CALC blocks are needed.
The first block calculates the intermediate variable Z:
   Z = I - X, where I = 100 The following formula depends on the value of the parameter L:
   if L <1, then
   Y = X / (Z * G + X), where G = 100L if L> 1 then
   Y = X / (Z / G + X), where G = 100 / L

Work with the linearization wizard

Ability to stretch or compress the image of the linearizer.
 The following options are available after selecting the Finish key in the Linearization Wizard.
Expand Top or Bottom - Stretch the top or bottom of the image.
The size of the image will increase so that you can see and move the blue markers - the ends of the linearizer curve.
Shrink Top or Bottom - Compress the top or bottom of the image.
These keys appear if the image has been stretched before. Click for more information.
Shrink All or Expand All - Compress or stretch everything.

With one click of this key, you can compress or stretch the entire image of the linearizer. The key appears in the center if the image has been stretched before this, and works in the toggle mode.
Notes. 1. The output of the regulator and the output of the linearizer are different variables. For example, if the final control element is a valve, then with a regulator output of 50%, the valve (after the linearizer) will not be in the 50% position.
2. If resetting feedback is used in the PID controller to suppress overshoot (Foxboro, Moore Products), and you have installed a linearizer in the loop circuit up to the Auto / manual station (apparently, this means a stand-alone controller outside DCS) , then feedback will require the use of an inverse linearizer.

Step 4. Selecting a linearizer programming language

Possible linearizer programming languages are presented for selection in the Linearization Wizard window at the top right. These are FORTRAN, Basic, C, Structured text, and in the case of a piecewise linear linearizer, there is also a table of X, Y coordinates of definition points of linear segments (X-Y pair list). On the graph, the linearizer input X is the right vertical axis, and the linearizer output going to valve control = horizontal Y axis.
The linearizer program in the wizard can be edited, copied or deleted. When copying and deleting the text of the program is written into the Windows buffer, where it can be inserted into any Windows application.


Step 5. Adjusting the regulator after linearization

XTune uses the collected data to perform all of its functions — analyzing, modeling, and loop configuration. These data should be identical to those that the regulator itself “sees”.
For an adjustable PV variable, always use the data that the controller uses for its PV. For CO, always use data directly from the output of the regulator. (See figure below)


For the output linearizer, always use the signal directly at the regulator output, but not at the linearizer output, to collect data.
In the case of a linearizer pH input, always use a linearized signal, that is, the PV and SP signals at the linearizer output (at the regulator input).


The asymmetry of the circuit is expressed in the dependence of the reaction of the process (load) on the direction of change in the output of CO: plus or minus, increase or decrease, with the same magnitude of change.
After the non-linearity test to check the asymmetry, it is necessary to perform several of the same steps in the manual mode of the regulator (open loop), which were done in the nonlinearity test, but in the opposite direction.
Analysis. Compare the resulting PID settings or mathematical process models presented in the Process Model simulation window. Does the reaction of the process when changing the CO up and down?
If the reaction is different, is it possible to eliminate or reduce this difference? If the situation cannot be rectified, it is necessary to use more conservative settings.


Data acquisition for integrating circuits

Some circuits are not capable of self-regulation because they contain an integrator, or there may be a large capacity in the loop. Such technological processes after a manual jump at the regulator output for a long and very long time (perhaps never) cannot achieve a stable state.
Examples are control loops:
• level of liquid product (in most cases)
• composition of reactants (composition) and temperatures of distillation columns and chemical reactors with a mixer
• extruder / press temperatures
• gas pressure (in some cases)
• digital mixing systems with volume control

There are 2 data acquisition methods for integrating circuits.

Method 1 - Stable condition. Use the standard data acquisition procedure for setting the contours described above in the appropriate section in Automatic or Manual controller operation mode. ExperTune program is designed to calculate the optimal settings, including integrating circuits. However, it may be very difficult to achieve a stable state in such a circuit, which is necessary for proper data collection. Use the following guidelines for this.
Recommendation 1. Automatic mode.
For a new loop or loop with cyclical oscillations, we recommend to collect data in automatic mode when adjusting the controller to proportional control only, that is, without integral and differential components. For stable control, set the proportionality factor of the controller (PG, P) to 0.5. If the integration time constant (I, TI) is measured in time / repeat units (for example, min / repeat), then set this parameter to the maximum possible value (in the APACS + / QUADLOG TImax = 4000.0 min / rep system). Without integration, the loop quickly comes to a “stable” state. If the inverse unit of rep / time is used, then set TI = 0.
After that, perform data collection with a change in the SP order and be sure to comply with all requirements.
 Recommendation 2. Manual mode.
To set up the integrating circuit in manual mode, we recommend the “quick” test described earlier in this manual. The test must begin and end with a stable state of the variables PV and CO.
If these recommendations are not appropriate or can not help, use the second method.
Method 2 - Permanent trend slope. If you are unable to achieve a stable state of the circuit, either in manual or automatic mode, use the method described in this section.
The idea of ​​this method is to collect data in manual mode, changing the output of the CO controller in a similar way to the “fast” test, and then use XTune to analyze the values ​​of the derivative of the adjustable variable PV.
A change in CO output leads to a change in the slope of the PV trend, that is, the rate of change of the adjustable variable PV. Do not change CO until the constant slope of the PV trend is established, that is, the constant value of the PV derivative. Then change the CO and wait again until the new constant slope of the PV trend is established. See the picture below.


After collecting data manually, press the “d / dt” key in the Trend window or select the Integrating loop in the Options menu. In this case, the contour is defined as “integrating”, and the XTune program will use the integrating process as a mathematical model of the technological process — the load of the contour. The setup wizard for integrating circuits is called up on the screen.

Integrated Circuit Configuration Wizard

To configure and analyze the integrating circuit, XTune calculates the derivative, that is, differentiates the signal of the adjustable variable PV. In addition, to suppress excessive noise of the PV signal, you must add a filter. Choose as small as possible (narrow) filter. However, it should be wider than double the data collection interval. By default, the filter time constant is set to 4 data collection intervals. To try different filters, use the Test key in the “Process variable filtering” filter installation panel.

The following figure shows the same collected data, but after differentiation.
Data after differentiation can be normally processed and analyzed, including the calculation of PID settings, statistical analysis, hysteresis testing, valve sticking, etc.

To adjust the PID controller using the transformed data, find and select in the Trend window using the Zoom In option a suitable area including trend sections with a stable state of the differentiated variable PV before and after changing the output of the CO controller (see figure below).


There are two main options for level control in the technological scheme with a liquid product reservoir and input and output streams:

1. You need normal precise level control according to the established SP task, your tank is a process tank, not a “buffer” tank. In this case, simply collect the data for tuning the loop and call the function to calculate the optimal PID settings (PID Tuner). The fastest way to collect data for a level controller is in automatic mode with tuning to purely proportional control without integral or differential action. See also the section “Data Collection for Integrating Circuits”.
2. You need to stabilize one of the streams, for example, the output, and the other stream and the level of product L in the tank can vary, but without overfilling or emptying the tank. In this case, the reservoir acts as a “buffer”, which smoothes out random disturbances, jumps in the load of the regulator, stabilizing the regulated input or output flow. The figure below shows the output flow stabilization scheme with unregulated input flow. For maximum efficiency of the buffer tank the level of the product in it should change, “float”. Then the adjustable flow rate is much less dependent on load spikes and other unstable flow rates. Such a scheme is called Averaging level control (Smoothing or "buffer" level control). In this scheme, there is no optimal value of L and the level set by the technology of the task.

Buffer (smoothing) level control

If your tank is designed to provide a more stable flow rate (flow) in the next section or process operation, then you may not need optimal or “hard” level control. In this case, you regulate the output flow, and the input flow may vary. The purpose of this “buffer” tank is to absorb, dampen changes in the input stream and smooth out and stabilize the output stream as much as possible, but without overflowing or emptying the reservoir.
Regulator tuning recommendation:
1. Use a proportional (P) -controller without integral and differential components.
2. Set the SP reference to the minimum acceptable Lmin level and designate it for process operators as the “lower limit”. Depending on the flow rate, the level in the tank will change (drift), and the name “task” may confuse the operators, since in reality this is the lower limit.
3. If your controller uses Proportional gain (PG) as a parameter of proportional gain / gain, set its value to:
100 / [(maximum permissible level in%) - (minimum permissible level in%)]
If the Proportional Band (PB) proportionality band is used, then PB = 100 / PG.
4. Set the parameter Bias (offset) = 0

The formula for PG assumes that the controller output is measured in%. When PG = 1 (PB = 100), a level change of 10% causes a change in the output of the CO controller also by 10%.

Such a regulator can also be used in a cascade control scheme as a leading (primary) regulator.


In such a system, at a high flow rate (flow rate), the level in tank L will be high, since it is more likely that the next disturbance will be in the direction of decreasing flow. Conversely, at a low flow rate, the L level will be low, since in this case the next change is likely to be aimed at increasing the flow.
If you need this type of regulation, use the XTune Wizard.
 Master Buffer Level Control
The ExperTune program has a dedicated “Buffer Level Regulation Wizard” to help the user go through the entire level controller setting procedure. The wizard is called up in the XTune window (PID Tuner) from the Options-Level wizard options menu.
First you need to decide what the tank is for and which level control option you need. “Strict” level control according to SP assignment? Or should the tank dampen flow changes?
If the task of the tank is to provide a more stable flow rate for the next technological operation or apparatus, then you need buffer control, and to continue work, select the item "As a buffer tank for smoothing flow surges) .
Otherwise, you do not need this Master, see option 1 in the previous paragraph.

There are several varieties (types) of such regulation in the Buffer Regulation Wizard. When working with the Wizard, windows are sequentially displayed on the screen in which the user makes a choice or enters additional information. Below are listed and described the main window, sufficient for the procedure of calculation and adjustment of the regulator.
First, the “Type of Averaging Level Control” window is displayed with information that helps to reasonably choose the type of buffer control

Type of Averaging Level Control - About the type of buffer level control

Although the “buffer” use of the reservoir is useful and desirable, but at the same time most of the process operators would like the setting of the SP regulator to be present in the level controller and the level would tend to return to this value. For this, the integral action of the regulator is necessary.

The advantages of using the integral component and setting the SP for the level:
• Operators - technologists, if they are not specifically retrained, it is more convenient to work with a level that returns to the set task.
• It is desirable that in case of equipment failure, for example, a pump shutdown, some time remains in stock.

Disadvantages of using the integral component:
• The volume of the tank, which can be used as a buffer, is reduced. For example, if the flow rate was maximum, and the SP target and level were 50%, then in the case of a disturbance causing a decrease in flow, the system has only half of the reservoir to compensate for this disturbance.
• Contour with integral action is more at risk of cyclic oscillations

Averaging Level Control with no Integral action - Buffer level control without integral component

The use of proportional control is selected. In this window, XTune calculates the proportionality coefficient P, PG for your controller.
Enter the values ​​in% for the upper and lower limits of the product level in the tank. Based on these values, the proportionality / gain coefficient (Gain) or the Proportional Band (PB) proportionality range is calculated depending on the type of controller or system, while PB = 100 / PG.
A simple calculation shows that the proportionality range of the PB regulator should be equal to the difference between the upper and lower limits of the level. Then the movement of the valve from the fully open to the fully closed state corresponds to the full range of allowable level change between the two limits.
The integral and differential components of the controller should not be used.
Set the SP value corresponding to the entered lower level limit.
For more on this topic, see the “Buffer Level Adjustment” section above.

Averaging Level Control with Integral action - Buffer level control with integral action. Calculate P.
You have chosen the option with an integral component, so that after a disturbance the level will slowly return to the SP task. In this window, the proportional component setting is calculated. Further in the following windows the integral component will be defined.

The proportionality coefficient depends on the parameters of the buffer capacity and the magnitude of the disturbance flow, which must be compensated. Enter the following data:
 Range of Manipulated flow: Enter the maximum range of controlled flow in engineering units.
• In a cascade. Often the output of the level controller in a cascade circuit is connected to the flow regulator. In this case, enter the maximum range of the flow regulator.
• In a single circuit. If there is no cascade and the output of the level control goes directly to the valve, enter a value equal to 1.3 of the estimated valve flow rate.

Distance from setpoint to nearest limit. The difference between the SP task and the closest (upper or lower) level limit, expressed in% level.

Normally expected flow uset. The expected value of flow disturbances in engineering units of the flow scale is normally expected. Check the correctness of the choice of flow units. They are used further to calculate the integral component.

Vessel volume known? Is tank volume known?
The value of the integral component for buffer level control depends on the time of filling the tank, that is, on its volume. If the volume is known, select “Yes” and enter this value in the next window. If not, the program will tell you how to measure it.

Enter vessel volume. Enter the tank volume.
Enter the buffer volume of the tank between the level limits (level taps). ExperTune uses this value of the volume and the previously entered value of the possible change (perturbation) of the flow rate to calculate the integral component.
Make sure the correct tank volume units are selected. The program compares them with the units of flow in the previous window. In case of inconsistency, a warning is issued, but if necessary, the program will convert units of volume. Next, the program calculates the parameter of the integral action in the units used by your regulator.

 Measure vessel volume. Measure the volume of the tank.
To start the measurement, it is necessary to set up the system so that the level in the tank does not change, while the level control must be in manual mode. To achieve this, it may be necessary to manually change the controller output up and down. Then, in this stable state, we change the flow rate by a known amount. By measuring the time it took for the resulting level change, the program calculates the volume of the tank and then the desired setting for the integral action of the controller.
Level measurement procedure.
1. When the level control loop is stable, move the controller to manual mode. In the case of a cascade control circuit, turn off the cascade and switch the secondary (internal) flow control loop to manual mode. Record the initial flow and level values.
2. Change the flow valve control signal by about 5%. In the cascade scheme, this will be the output of the flow controller of the secondary circuit. In the single loop circuit, this will be the output of the level control.
3. Wait for the level to change by about 5% and record the new flow and level values, as well as the elapsed time. To avoid problems with the process, return the flow valve to its original position and turn the circuit (or cascade circuits) back to automatic mode.

In the same window, enter the results of 5 such measurements in the designated fields. ExperTune first calculates the volume of the tank. If the volume value looks wrong, check the measurements made and the engineering units of the flow. In case of erroneous units of consumption, go back to the window of calculating the proportionality coefficient of the regulator and change the unit of measurement for the flow.
ExperTune uses the found value of the volume and the entered value of the possible change (perturbation) of the flow rate to calculate the recommended parameter of the integral component of the controller.
This parameter will be presented in the units accepted in your regulator. For the APACS + / QUADLOG system, this is the integration time constant TI (in ExperTune denoted by I), measured in units of min / repeat (min / repeat)

Level as a Surge vessel .. Level as a buffer tank.
The program shows the recommended values ​​of the Proportional (P) and Integral (I) actions for buffer level control by your loop. In this case, it is always necessary that the differential component is completely absent (D = 0).
 The recommended value of the parameter I = 0 means the absence of an integral component. In some types of regulators that use units of time - seconds or minutes to measure I, to exclude the integral action, a large number must be entered (in the APACS + / QUADLOG system, TImax = 4000 min).


In processes with an inverse (opposite) reaction, the Inverse Response Process, when the output of the CO controller changes, the adjustable variable PV begins to move “incorrectly” - in the opposite direction from the SP reference, not in the same way as the regulator sets (see figure below). These processes are also referred to as shrink-swell processes.
The option to configure the control loop of such a process is called up in the Edit Setup window: the Advanced - Loop Setup ... - Advanced-Inverse response process key.
The process with back reaction is a process that has a lead link with the “advance” of the input signal and a negative value of the time constant Lead time. In practice, this type of circuit usually also has an integrator - integrator or a large delay time Lag time. In ExperTune, the Inverse Response Process option assumes that the process has the following properties: Lead time negative, integrator link (or long Lag time), as well as lag link delay and dead time Dead time. These processes are second order processes.
  Two examples of such processes:
• Regulation of the level in the boiler drum
• Product level control at the bottom of the distillation column

Use this option only if you are sure that you are dealing with a process of this particular type.

If the option “Process with backlash” is selected, the program analyzes the data and finds the process model under the assumption that it includes the lead, lag and integrator links, automatically determines the “negative” burst parameter (lead) and the integration time. The PID program settings will be specifically targeted to this type of process.


For each control loop, ExperTune generates a detailed sample report with analysis and loop setup results. If you later encounter any problems with the contour, you can always use this report for comparison.
The report generation function can be called from the options menu in the faceplate window: Options - Tuning Report or from the “Data Data for ...” Data Trends Window: Options - Report - Full Tuning Report.
In addition, all ExperTune windows have an Add to Report option in the Options menu.
After adding new data to the report, you must wait until the hour is complete (the hourglass icon should disappear), without loading the computer with anything else. After the report is generated, it is necessary to update all its fields. To do this, from the MS Word Edit menu, select Select All and press the F9 key.

File - Report Template
ExperTune creates reports in MS Word 97 or higher, therefore report files are .doc files. Reports are created based on a generic generic .doc template file that contains an “empty” report structure. By default, this file is located in the ExperTune / Common folder. In the Advanced Options window — on the Report tab, the user can select a different folder and another file for the report template (see the figure below). The window is called from the ExperTune start window via the options menu: Options-Advanced.
There may be several report templates, for example, for different categories of control loops.
The user can edit the report template according to customer requirements.
When the function of the report is called for the first time (for this contour), for example, Add to report, ExperTune copies the LoopReport.doc file to your folder where
.tun files and archive files of the contour. The report file receives the same root name as the mentioned files, plus the extension .doc. If you have already caused a report, the program asks if you want to use an existing report or replace it with a new template.     

Reporting method
Reports are created using bookmarks in a Word document. ExperTune adds the desired information to the place of these bookmarks in the report. Each block of information in the report template has its own tab. Three bookmark categories are used: Graph (Graphics), Picture (Picture) and Text (Text) for three types of data blocks of the control loop. Each block of information placed in a bookmark begins with text indicating the type of this block: Graph, Picture or Text.
The “Graph” type tab contains a Word Picture graphic object for placing an image of the ExperTune window with trend graphs (such as the “Time data for ...” data trend window or the “Robustness plot” contour stability graph). Bookmark contains only one such object and nothing else.
Bookmarks of the "Picture" type are intended for windows and panels with numerical and textual information and dialogue tools (calling and entering data, choosing an option in the list, etc.), but without graphic trends. These tabs contain a single Word Picture object that defines the size and location of the corresponding block of Picture information. An example is the online statistical analysis panel “Statistical Analysis”.
Bookmarks that begin with the word "Text" contain text that is replaced by the program when creating a report for a particular outline. The user cannot create his own bookmarks of this type.

The table below shows the prefixes / prefixes (beginning of the name) of the bookmarks used by ExperTune to include in the report various blocks of contour information.

When ExperTune receives a command to add information to a report, it looks for a bookmark in the report.
with the appropriate name. If for information such as Graph and Picture there are two or more suitable bookmarks, then the program issues a message and asks the user to select a bookmark. Any object with a bookmark can be moved and resized. You can also add your own bookmarks like Graph and Picture, but you cannot add your own bookmarks like Text ..
 Note. Do not delete bookmarks from the document template, otherwise this part of the report will not work.
You cannot change existing bookmark names.

User change report template
ExperTune's standard report template can be edited and edited to fit a particular company, individual circuit, or category of circuit. For example, if there are several types of contours, then you can make a separate template for each such group.
Reports are created using bookmarks in a Word document. To add your own bookmarks in Word, first define the object for which you want to create a bookmark, and then in the Word environment, select Insert> Bookmark. Next, use the rules for generating bookmark names described above in the Report generation method section.

Bookmarks for information objects such as “Graph” and “Picture” should contain, respectively, only one window with a graphic image or one dialog box. ExperTune will use user-defined dimensions for these objects.
A bookmark for text can contain any number of characters.

This macro adds comments to all the bookmarks in your document, with each added comment beginning with the text "BookMark:". In order for comments to work in Word as "pop-up" screen prompts, select the Tools menu (Tools)> Options (Options), then click on the View tab and select "Tooltips".
To run a macro in Word, go to Tools> Macro> Macros (Tools-Macros-Macros).
Since the reports contain macros and a program on Visial Basic (VBA), when opening a report or report template in the Word environment through the File> Open menu, Word may issue a security warning message: "The document contains macros in which there may be viruses harmful to your computer ... " To continue, click the “Enable Macros” button.

#Optimization, #ExperTune, #Cycle, #processing, #Zone, #insensitivity, #Compensation, #proactive, #feedforward, #regulation, #Unstable, #processes, #tuning, #coefficients, #PID, #Frequency, #analysis , #settings, #cascading, #contours, #calculation, #Performance, #increase, #efficiency, #Setpoint

Be the first to comment

You comment add

Back to the list