CONFIGURATION GUIDE. CG39ARCAL-1 Rev: 1 January APACS+ TM Advanced Regulatory Control Application Library Version 4.

Transcription

1 CONFIGURATION GUIDE CG39ARCAL-1 Rev: 1 January 1999 APACS+ TM Advanced Regulatory Control Application Library Version 4.00 or Higher

2 CONTENTS TABLE OF CONTENTS SECTION AND TITLE PAGE 1.0 INTRODUCTION PRODUCT DESCRIPTION PRODUCT SUPPORT RELATED LITERATURE SINGLE_LOOP_FF (Single Loop with Feedforward) TUNING SINGLE_LOOP_AG (Single Loop with Adaptive Gain) TUNING SINGL_LP_ONOFF (On/Off Single Loop) TUNING SINGLE_LOOP_SR (Split-Range Single Loop) SINGLE_LOOP_TP (Single Loop with Time-Proportioned Discrete Output) SINGLE_LOOP_FB (Single Loop with External Feedback) TUNING SINGLE_LOOP_OR (Single Loop with Override Selector) TUNING SINGLE_LP_B_SW (Single Loop with Batch Switch) TUNING SINGLE_LP_B_EXT (Single Loop with Batch Extension) TUNING SINGLE_LP_PRG (Single Loop with Programmed Setpoint) SINGLE_LOOP_MD (Single Loop with Maximum Deviation) SINGLE_LP_STPWT (Step and Wait Single Loop) TUNING ANALYSIS OF STEP AND WAIT CONTROLLER SINGLE_LP_CMPFB (Complimentary Feedback Single Loop) TUNING ANALYSIS OF COMPLEMENTARY FEEDBACK CONTROLLER January 1999 i

3 CONTENTS CG39ARCAL SINGLE_LP_SM_PR (Smith Predictor Single Loop) TUNING ANALYSIS OF SMITH PREDICTOR CONTROLLER SINGLE_LP_CF_SR (Split-Range Coarse/Fine Single Loop) SINGLE_LP_CF_FL (Floating Coarse/Fine Single Loop) TUNING SINGLE_LP_CF_CS (Center-Seeking Coarse/Fine Single Loop) TUNING BAL_2FE (Balance of Two Final Control Elements) BAL_3FE (Balance of Three Final Control Elements) BIAS_FE (Bias of a Final Control Element) INTERP_XY (Interpolation of X-Y Data) INTERP_XYZ (Interpolation of X-Y-Z Data) ST_TIME_SLICE_EX (Structured Text Time-Slice Example) LIST OF TABLES FIGURE AND TITLE PAGE 1-1 TIC CONTACT INFORMATION Moore Products Co. assumes no liability for errors or omissions in this document or for the application and use of information included in this document. The information herein is subject to change without notice. The Moore logo, APACS+ and 4-mation are trademarks of Moore Products Co. All other trademarks are the property of the respective owners. Copyright 1999 Moore Products Co. All rights reserved ii January 1999

4 INTRODUCTION 1.0 INTRODUCTION The Advanced Regulatory Control Application Library (Version 4.00 or higher) is a set of derived function blocks (DFBs) that are pre-configured at the factory to perform advanced regulatory control functions within the framework of the 4-mation TM configuration software. The library is in the form of a stand-alone, off-line, advanced control module (ACM) configuration that is part of a system named BATCH. The configuration is opened like any other off-line database. When open, the blocks can be selectively copied from the library configuration and pasted into a user-developed configuration. The library can be used directly from the floppy or from a hard disk after first copying the entire library floppy contents (with all subdirectories intact) onto the hard disk. This configuration guide presents support information for understanding and using the function blocks of the Advanced Regulatory Control Application Library. This information includes a description of what each block does, the inputs and outputs of the block, and how to tune the blocks where applicable. The guide is organized with a separate section for each function block in the application library. The sections are presented in the following order: Section 2, SINGLE_LOOP_FF (Single Loop w/ Feedforward) Section 3, SINGLE_LOOP_AG (Single Loop w/ Adaptive Gain) Section 4, SINGL_LP_ONOFF (On/Off Single Loop) Section 5, SINGLE_LOOP_SR (Split-Range Single Loop) Section 6, SINGLE_LOOP_TP (Single Loop w/ Time-Proportioned Discrete Output) Section 7, SINGLE_LOOP_FB (Single Loop w/ External Feedback) Section 8, SINGLE_LOOP_OR (Single Loop w/ Override Selector) Section 9, SINGLE_LP_B_SW (Single Loop w/ Batch Switch) Section 10, SINGL_LP_B_EXT (Single Loop w/ Batch Extension) Section 11, SINGL_LP_PRG (Single Loop w/ Programmed Setpoint) Section 12, SINGLE_LOOP_MD (Single Loop w/ Maximum Deviation) Section 13, SINGL_LP_STPWT (Step and Wait Single Loop) Section 14, SINGL_LP_CMPFB (Complementary Feedback Single Loop) Section 15, SINGL_LP_SM_PR (Smith Predictor Single Loop) Section 16, SINGL_LP_CF_SR (Split-Range Coarse/Fine Single Loop) January

5 INTRODUCTION CG39ARCAL-1 Section 17, SINGL_LP_CF_FL (Floating Coarse/Fine Single Loop) Section 18, SINGL_LP_CF_CS (Center-Seeking Coarse/Fine Single Loop) Section 19, BAL_2FE (Balance of Two Final Control Elements) Section 20, BAL_3FE (Balance of Three Final Control Elements) Section 21, BIAS_FE (Bias of a Final Control Element) Section 22, INTERP_XY (Interpolation of X-Y Data) Section 23, INTERP_XYZ (Interpolation of X-Y-Z Data) Section 24, ST_TIME_SLICE_EX (Structured Text Time-Slice Example) 1.1 PRODUCT DESCRIPTION This application library consists of derived and standard function blocks that perform advanced regulatory control functions such as Feedforward, Batch Switch, Coarse/Fine, and Smith Predictor. Additional control features are provided by the balancing circuits for multiple final control elements and an auto/manual biasing circuit for a final control element. The library also includes two data interpolation blocks, one for Y vs. X data (an expandable linear characterizer) and one for Z vs. X and Y data (a table lookup block), and a structured text block that demonstrates how a complex calculation can be time-sliced (executed in parts) to minimize the impact on each individual controller scan. 1.2 PRODUCT SUPPORT Product support can be obtained from a Technical Information Center (TIC). Each regional TIC is a customer service center that provides direct telephone support on technical issues related to the functionality, application, and integration of all products supplied by Moore. Regional TIC contact information is provided in Table 1-1. Your regional TIC is the first place you should call when seeking product support information. When calling, it is helpful to have the following information ready: Caller ID number, or name and company name - When someone calls for support for the first time, a personal caller number is assigned. Having the number available when calling for support will allow the TIC representative taking the call to use the central customer database to quickly identify the caller=s location and past support needs. Product part number or model number and version If there is a problem with the product=s operation: - Whether or not the problem is intermittent - The steps performed before the problem occurred - Any error messages or LED indications displayed - Installation environment 1-2 January 1999

6 INTRODUCTION Customers that have a service agreement (ServiceSuite or Field Service Agreement) are granted access to the secure area of our Web site ( This area contains a variety of product support information. To log on, you will be prompted to enter your username and password. TIC North America also offers a free faxback service called FaxRequest. You can dial-in to this service to access documents such as press releases, product information sheets, and training schedules. The service is completely automated and available 24 hours a day. To access this service, call the FaxRequest number listed in Table 1-1. The first document you should request is the directory (document number 9999). This document is constantly updated as new documents are added. Each document has a number code assigned to it that you enter along with your fax number (area code entry is always required). Upon completing your entry, the FaxRequest computer automatically calls your fax machine and sends the requested documents. TABLE 1-1 TIC CONTACT INFORMATION Tel: , extension 4842, option 1 Fax: TIC NORTH AMERICA natic@mpco.com FaxRequest: , extension 4842, option 2 Bulletin Board Service: Hours of Operation: Secure Web Site: 8 am to 6 pm eastern time Tel: TIC ASIA-PACIFIC Fax: aptic@mpco.com Hours of Operation: Secure Web Site: 9 am to 6 pm Singapore time January

7 INTRODUCTION CG39ARCAL-1 TABLE 1-1 Continued Tel: TIC EUROPE Fax: Hours of Operation: Secure Web Site: 8:30 am to 5:15 pm GMT/BST RELATED LITERATURE The following Moore literature is available for reference: AGA Natural Gas Flow Calculations in the APACS Controller (AD39-3) For 4-mation configuration software, versions 3.XX: - 4-mation User s Manual, Installation and Operation (UM39-6) - 4-mation User s Manual, Function Block Languages (UM39-7) For 4-mation configuration software, versions 4.XX: - 4-mation Installation and Operation (UM39-11) - 4-mation Function Block Language (UM39-12) The following vendor literature should be available as needed: Microsoft MS-DOS Operating System Reference Microsoft Windows 3.1 (or later) Operating System Reference 1-4 January 1999

8 SINGLE_LOOP_FF 2.0 SINGLE_LOOP_FF (Single Loop with Feedforward) The derived block SINGLE_LOOP_FF is a steady-state, feedforward, single loop controller. It is a modified SINGLE_LOOP Basic block with a feedforward value added to the output of the controller block. With this modification, the output changes immediately when the feedforward variable changes and the controller block merely trims the feedforward value to ensure that the process variable equals the setpoint. The FF_LEADLAG block allows dynamics to be applied to the feedforward variable. The GAIN softlist value of the FF_LEADLAG block scales the feedforward variable. The resulting value, FF, is added on to the controller output by the FF_ADD block. For the controller to operate properly, the feedback signal must be comprised of only the controller s contribution to the output, so the FF value is subtracted from the [] value at the FF_SUB block. Note that the output range of the CONTROLLER block is -100 to 100 (not 0 to 100) so that it can trim the feedforward variable over the entire range. The inputs and outputs for the SINGLE_LOOP_FF block include: FF_ = process variable = feedforward variable = setpoint = output 2.1 TUNING The controller block should be tuned first. The feedforward variable should be held at a fixed value while the controller is tuned using standard methods. Next the feedforward variable is tuned. The appropriate gain and lead or lag times are determined to minimize the change in controlled variable for a change in load (feedforward) variable (with the controller in automatic mode). For more information on feedforward control, see Feedforward Control Using The Model 352 Single- Loop Controller, AD January

9 SINGLE_LOOP_AG 3.0 SINGLE_LOOP_AG (Single Loop with Adaptive Gain) The derived block SINGLE_LOOP_AG is a single loop controller with adaptive gain. It is a modified SINGLE_LOOP Basic block with the controller gain automatically adapted based on the value of the process variable. With this block, the base PG value is multiplied by either a 1.0 or a 0.5 before it is written to the controller as the actual proportional gain. The PG block holds the base value of the controller proportional gain value. Note that, when tuning, the proportional gain changes must be made to the PG block (not the CONTROLLER block PG softlist parameter). The AG_SEL block and the AG_MUL block adapt the base PG value according to the value of the process variable, and the PG_SET_VAL block continuously writes the resulting value to the PG softlist parameter of the CONTROLLER block. The input and outputs of the SINGLE_LOOP_AG block include: = process variable = setpoint = output 3.1 TUNING The controller block should be tuned using standard methods. Again, note that the base proportional gain value in the PG block is used for tuning, not the CONTROLLER block PG parameter. January

10 SINGL_LP_ONOFF 4.0 SINGL_LP_ONOFF (On/Off Single Loop) The derived block SINGL_LP_ONOFF is an on/off single loop controller. It is a modification of the SINGLE_LOOP Basic block with the PID controller replaced by an on/off controller that determines the value of a discrete output. The CONTROLLER block is an ON_OFF block that evaluates the and inputs and determines if the discrete output(s) should be energized. The ON_OFF block can be configured in a variety of ways; in this example, the L and H outputs are used with the _RS flipflop to determine the value of one discrete output. The DEADBAND block holds the deadband value, which represents an acceptable difference between and. When the is less than by more than the deadband value, the output is energized and will remain energized until the has been raised above the by an amount equal to the deadband. The AM_SEL block selects the calculated output in auto mode ([] is TRUE), and allows it to be changed in manual mode ([]=FALSE) via a write to []. On a transition from auto to manual mode, the output is de-energized (via the _F_TRIG and _MOVE blocks). The inputs and outputs of the SINGLE_LP_ONOFF block include: = process variable = setpoint = discrete output 4.1 TUNING In this example, the deadband is the only value to be tuned. January

11 SINGLE_LOOP_SR 5.0 SINGLE_LOOP_SR (Split-Range Single Loop) The derived block SINGLE_LOOP_SR is a split-range, single loop controller. It is a modification of the SINGLE_LOOP Basic block with the controller output value converted into two output values. One output value increases the process variable while the other output value decreases it, but only one output variable is active at a time. A deadband value is used to create a gap around the 0 crossover point; inside the gap neither output is active. The CONTROLLER block output has the range -100 to 100, with 0 as the crossover point (the center of the gap where neither output is active). The DEADBAND block holds the value that forms the gap. If the difference between [] and 0 is less than the deadband value, both of the outputs are calculated to be no greater than 0. Calculations for each output, displayed in cells O5 and O7, result in an output of 0 at the edge of the gap and 100 at the end of the output range (100 or -100). The inputs and outputs of the SINGLE_LOOP_SR block include: = process variable _H _L = setpoint = controller output = output that is active when is high = output that is active when is low January

12 SINGLE_LOOP_TP 6.0 SINGLE_LOOP_TP (Single Loop with Time-Proportioned Discrete Output) The derived block SINGLE_LOOP_TP is a single loop controller with a time-proportioned discrete output. It is a modification of the SINGLE_LOOP Basic block with the analog output replaced by a discrete output. The discrete output acts as a modulating output by varying the percentage of a timed cycle during which the output is energized. The _REPCYCL block converts the [] controller output into the [D_] discrete output value. The ONTIME softlist parameter of the _REPCYCL block is the duration of the cycle, and the PCTON input (the [] value) is the percentage of the cycle during which the output will be energized. Be aware that the output cannot change more frequently than the controller scan rate, so the effective resolution of the output will depend on both the scan rate and the cycle time. The inputs and outputs of the SINGLE_LOOP_TP block include: = process variable D_ = setpoint = controller output = discrete output January

13 SINGLE_LOOP_FB 7.0 SINGLE_LOOP_FB (Single Loop with External Feedback) The derived block SINGLE_LOOP_FB is a single loop controller with external feedback. This block can serve as the override loop in an override control scheme (see the SINGLE_LOOP_OR block in Section 8.0). The block s only modification of the SINGLE_LOOP Basic block is the source of the CONTROLLER block s FDBK feedback input. Instead of [], it is the block s [FDBK] external feedback input that has the value of the selected output in an override scheme. The inputs and outputs of the SINGLE_LOOP_FB include: FDBK = process variable = external feedback (the selected output in an override scheme) = setpoint = output 7.1 TUNING The controller block should be tuned using standard methods. Be sure that the other loops in the override scheme are never selected while tuning this loop; this can be accomplished by disabling the other controllers outputs and holding them at a value that cannot be selected. For more information on override control, see Override Control Using The Model 352 Single-Loop Controller, AD January

14 SINGLE_LOOP_OR 8.0 SINGLE_LOOP_OR (Single Loop with Override Selector) The derived block SINGLE_LOOP_OR is a single loop controller with an override selector. It receives override values from other sources, such as the SINGLE_LOOP_FB override loop (see Section 7.0). It is a modification of the SINGLE_LOOP Basic block with a low selector placed downstream of the controller block to enable an override input to be selected over the controller output. The OVERRIDE_SEL block selects the minimum between the control loop output and the [OR_IN] override input (probably from another control loop). The OS_NOT block determines the [OVRRD] override status output. If the control loop output is not the selected output, [OVRRD] becomes TRUE. Note that the OVERRIDE_SEL block is downstream of the _MANUAL block, enabling the override input to be selected over the operator s manual setting. If you want the operator to always be able to set the output in manual mode, then the order of the _MANUAL and OVERRIDE_SEL blocks should be reversed. The inputs and outputs of the SINGLE_LOOP_OR block include: OR_IN = process variable = override input (probably from another control loop) OVRRD = setpoint = output = override status 8.1 TUNING The controller block should be tuned using standard methods. Be sure that the other loops in the override scheme are never selected while tuning this loop. To de-select the loops, disable the outputs of other controllers and hold them at a value that cannot be selected. For more information on override control, see Override Control Using The Model 352 Single-Loop Controller, AD January

15 SINGLE_LP_B_SW 9.0 SINGLE_LP_B_SW (Single Loop with Batch Switch) The derived block SINGLE_LP_B_SW is a single loop controller with a batch switch, a modified version of the SINGLE_LOOP Basic block with the batch switch block placed in the feedback line of the controller. The batch switch is used to prevent reset windup, which can occur when a controller is unable to bring the process variable to setpoint for an extended period of time (e.g. a temperature controller at the start of a batch). Reset action will drive the controller s output and its reset component beyond the end of its range (by a few percent). The controller output cannot change until the process variable crosses setpoint, at which point the output is still at the end of its range and will be late in responding, causing an overshoot. The batch switch, placed in the feedback line to the controller, will modify the feedback to the controller when the output is driven to the end of its range. The result is that the controller s reset component will not be able to wind up, and the controller will unwind much quicker when the process variable crosses setpoint. Note that the batch switch has no effect when the controller output is within its range (not wound up). The inputs and outputs of the SINGLE_LP_B_SW block include: = process variable = setpoint = output 9.1 TUNING The controller block should be tuned using standard methods. When tuning for reset windup prevention, the BPL (batch preload) softlist value of the batch switch can be changed. If the batch switch enables the controller to respond too quickly from a wind up (i.e. if the controller overshoots when unwinding), then the value can be moved away from the limit that the controller reached. If the controller responds too slowly from a wind up, then the value can be moved toward the limit that the controller reached. January

16 SINGLE_LP_B_EXT 10.0 SINGLE_LP_B_EXT (Single Loop with Batch Extension) The derived block SINGL_LP_B_EXT is a single loop controller with batch extension algorithm. It is a modification of the SINGLE_LOOP_SS Basic block with the controller output capable of being overridden by the batch extension algorithm. The batch extension algorithm is the fastest way for a controller to make the process variable equal the setpoint. The algorithm assumes that there are no restrictions regarding the speed at which the process variable changes or the difference between the process variable and the setpoint. It is a four step program that runs in the following order: 1. It sets the output to the maximum value until the process variable is within a deadband value of the setpoint. 2. It sets the output to the minimum value. 3. It holds the output at the minimum value for a set length of time (for the process variable to coast into the setpoint). 4. It sets the output to a configured value and releases the output for the controller to manipulate normally. The BAT_EXT_PGRMR block performs the batch extension program. The program is initiated when the block writes a TRUE to the BAT_EXT_EN variable while the controller is in auto mode. The steps of the program occur as follows: 1. Output set to 100 This step runs until the process variable [] is within the deadband value of the setpoint [], as determined by the SEG2_AND block. The deadband value is entered in the BAT_EXT_DBAND block. The SEG2_R_TRIG block creates the pulse which forces the program into the second step. 2. Ouput set to 0 3. Output held at 0 This step has a configured time length (the TIME_SEG_3 softlist parameter) for allowing the to coast up to the. 4. Ouput set to a configured value The configured value is the ENDPNT_SEG_4 softlist parameter When the program is complete, the DONE_R_TRIG and DONE_MOVE blocks reset the BAT_EXT_EN variable, allowing the controller to manipulate the output normally. The MAN_MOVE block ensures that the batch extension program cannot run in manual mode. NOTE An active batch extension program can be canceled by writing a FALSE to the BAT_EXT_EN variable. = process variable = setpoint = output January

17 SINGLE_LP_B_EXT CG39ARCAL TUNING The controller block should be tuned using standard methods. When tuning the batch extension algorithm, the values to be determined (probably by experimentation) are the: Deadband used to determine when to zero the output (the BAT_EXT_DBAND block s VALUE softlist parameter) Coast time while the output is zero (the BAT_EXT_PGRMR block s TIME_SEG_3 softlist parameter) Initial output value for the return to auto mode (the BAT_EXT_PGRMR block s ENDPNT_SEG_4 softlist parameter) January 1999

18 SINGLE_LP_PRG 11.0 SINGLE_LP_PRG (Single Loop with Programmed Setpoint) The derived block SINGL_LP_PRG is a single loop controller with a programmed setpoint. It is a modification of the SINGLE_LOOP Basic block with the setpoint that can optionally follow a program (e.g. a batch temperature profile). The PROGRAMMER block performs the setpoint program. Writing a TRUE to the [EXT] variable (while the controller is in auto mode) initiates the program. The PGRMR block accommodates up to 20 steps, and multiple PGRMR blocks can be cascaded if needed. To cancel a program, the [EXT] variable should be reset. Writing a TRUE to the [HOLD] variable will hold the sequence; the setpoint can then be changed (via the SETPOINT block) while the program is held and the program will resume at the new setpoint value. The DONE_R_TRIG, DONE_MOVE, EXT_MOVE, and HOLD_MOVE blocks reset the program variables when the program is complete or the mode is inappropriate (i.e. manual mode or internal mode). The inputs and outputs of the SINGL_LP_PRG block include: = process variable EXT HOLD = setpoint = output = external (program) status = program hold status January

19 SINGLE_LOOP_MD 12.0 SINGLE_LOOP_MD (Single Loop with Maximum Deviation) The derived block SINGLE_LOOP_MD is a single loop controller with a limit placed on the deviation between process variable and setpoint. It is a modification of the SINGLE_LOOP Basic block with extra limiting placed on the setpoint. This limiting acts to slow down a controller while a large error exists, such as at the start of a batch reaction. The MAX_DEV block holds the maximum - deviation. The _DEV_LIMIT block applies limiting to the setpoint which cannot differ from the process variable by more than the maximum deviation value. The inputs and outputs of the SINGL_LOOP_MD block include: = process variable = setpoint = output January

20 SINGLE_LP_STPWT 13.0 SINGLE_LP_STPWT (Step and Wait Single Loop) The derived block SINGL_LP_STPWT is a step and wait single loop controller, which is used for processes dominated by dead time. It is a modification of the SINGLE_LOOP Basic block with timing and tracking blocks added to perform the step and wait function. Dead time is the period of time during which the process variable does not respond to a change in output. This delay can cause poor control performance. For example, a PI controller makes immediate output changes through proportional action, which may be the exact change required, but the dead time delay prohibits the integral action from recognizing that the error is gone. The integral action then changes the output inappropriately. In the case where dead time is the dominant dynamic element in the control loop, some type of dead time compensation is required to improve control loop performance. Although dead time is present to some degree in nearly all process control loops, the dynamic behavior of most loops is dominated by one or more capacity lags in series with the dead time. As an approximation for the case where there are multiple lags, the process can be characterized by an effective dead time (θ), an effective lag time constant (τ), and a steady-state gain (K P ). Dead time compensation should be considered whenever the ratio of dead time to lag time constant (θ/τ) exceeds 0.5. The step and wait controller mimics the actions of an experienced control room operator who is manually controlling a process dominated by dead time. Whenever it is necessary to move the valve, the operator makes a valve change (the step) and waits for the process to respond completely to that valve change before making another step. The tuning of the step and wait controller is based on the process characteristics (θ, τ, and K P ). The effective lag time constant (τ), in min, is entered into the LAG_TIME block. The effective dead time (θ), in min, is entered in the DEAD_TIME block. The inverse of the normalized steady-state gain (1/K P ) is entered in the CONTROLLER block as the PG softlist parameter (note that K P is dimensionless since it is based on normalized process variable and output values). As an approximation, a lag response can be assumed to be complete (returned to steady-state) after four time constants (4τ). The LAG_MUL, DT_MUL, and DT_ADD blocks calculate the 4τ and 4τ+θ values and the TIME_CONVERT block converts the values to the TIME data type. A threshold value is entered in the THRESHOLD block. If the error between and exceeds the threshold value, the control circuit is enabled (by the ERR_ABS, ERR_GT, and ON_DELAY blocks). The ON_DELAY block, with the delay time set at 4τ, ensures that the full magnitude of the control error can develop before the first step occurs. If the error persists, subsequent steps will be determined by the REPEAT_CYCLE block. The REPEAT_CYCLE block has a minimal ONTIME softlist setting and the OFFTIME is set, via the OFFTIME_SET_VAL block, equal to 4τ+θ. This ensures that the process has responded completely to the last step before performing the next step. The REPEAT_CYCLE output is converted into a pulse, via the R_TRIG block, which becomes the momentary track command for the _TRK_HLD block, thereby passing the controller output to the _MANUAL block to perform a step. The TRK_OR block ensures proper tracking in manual mode. January

21 SINGLE_LP_STPWT CG39ARCAL-1 The inputs and outputs of the SINGL_LP_STPWT block include: = process variable = setpoint = output 13.1 TUNING First the lag time (τ) and dead time (θ) values are set (in minutes), then the optimum controller gain (PG) is set. The integral and derivative parameters are set at their minimum values. A step response test will produce the lag time and dead time values as well as the steady-state gain of the process. The optimum PG setting is the inverse of the normalized steady-state gain of the process (1/K P ). NOTE K P is dimensionless since it is based on normalized process variable and output values. If the PG is correct, it is possible to correct for a disturbance in one complete step and wait cycle. If the gain is too low, more than one step cycle is required. If the gain is too high, the will cycle about the ANALYSIS OF STEP AND WAIT CONTROLLER The step and wait controller provides a very robust method of dead time compensation. It is relatively easy to identify the process parameters required. Once the time parameters are set, tuning is reduced to a one knob (PG) tuning problem. The other dead time compensation methods presented in this library, Complementary Feedback (see Section 14.0) and Smith Predictor (see Section 15.0), are less robust but have the potential to perform better than the step and wait controller. For more information on step and wait control, see Dead Time Compensation Using The Model 352 Single-Loop Controller, AD January 1999

22 SINGLE_LP_CMPFB 14.0 SINGLE_LP_CMPFB (Complementary Feedback Single Loop) The derived block SINGL_LP_CMPFB is a complementary feedback single loop controller, which is used for processes dominated by dead time. It is a modification of the SINGLE_LOOP Basic block with the controller feedback delayed by a dead time block. Dead time is the period of time during which the process variable does not respond to a change in output. This delay can cause poor control performance. For example, a PI controller makes immediate output changes through proportional action, which may be the exact change required, but the dead time delay prohibits the integral action from recognizing that the error is gone. The integral action then changes the output inappropriately. In the case where dead time is the dominant dynamic element in the control loop, some type of dead time compensation is required to improve control loop performance. Although dead time is present to some degree in nearly all process control loops, the dynamic behavior of most loops is dominated by one or more capacity lags in series with the dead time. As an approximation for the case where there are multiple lags, the process can be characterized by an effective dead time (θ), an effective lag time constant (τ), and a steady-state gain (K P ). Dead time compensation should be considered whenever the ratio of dead time to lag time constant (θ/τ) exceeds 0.5. The complementary feedback controller modifies the reset feedback signal to a conventional PI controller. A dead time element is inserted in the reset feedback path to delay the integral action while the effect of the proportional action is delayed by the process dead time. The tuning of the complementary feedback controller is based on the process characteristics (θ, τ, and K P ). The effective dead time (θ), in min, is entered in the FDBK_DELAY block s DT softlist parameter. The effective lag time constant (τ), in min, is entered into the CONTROLLER block as the TI softlist parameter. One half of the inverse of the normalized steady-state gain (0.5/K P ) is entered in the CONTROLLER block as the PG softlist parameter (note that K P is dimensionless since it is based on normalized process variable and output values). The inputs and outputs of the SINGL_LP_CMPFB block include: = process variable = setpoint = output January

23 SINGLE_LP_CMPFB CG39ARCAL TUNING Tuning involves first entering the dead time value (θ) in the FDBK_DELAY block. Next, the lag time constant τ (in min) is entered as the controller integral time (TI) and then the optimum controller gain (PG) is set. The derivative parameter is set at its minimum value (0). A step response test will produce the lag time and dead time values as well as the steady-state gain of the process. The optimum PG setting is one half the inverse of the normalized steady-state gain of the process (0.5/K P ). NOTE K P is dimensionless since it is based on normalized process variable and output values. For more damping, the PG should be decreased from the optimum setting ANALYSIS OF COMPLEMENTARY FEEDBACK CONTROLLER The complementary feedback controller depends on an accurate estimate of the process parameters, but it is still fairly easy to tune. It is less robust than the step and wait controller (see Section 13.0), but more robust than the Smith Predictor (see Section 15.0). It has the potential to perform better than the step and wait controller, but the Smith Predictor has the potential to surpass its performance. For more information on complementary feedback control, see Dead Time Compensation Using The Model 352 Single-Loop Controller, AD January 1999

24 SINGLE_LP_SM_PR 15.0 SINGLE_LP_SM_PR (Smith Predictor Single Loop) The derived block SINGL_LP_SM_PR is a Smith Predictor single loop controller which is used for processes dominated by dead time. It is a modification of the SINGLE_LOOP Basic block where the process variable seen by the controller is modified by a model of the process to remove the effects of the dead time. Dead time is the period of time during which the process variable does not respond to a change in output. This delay can cause poor control performance. For example, a PI controller makes immediate output changes through proportional action, which may be the exact change required, but the dead time delay prohibits the integral action from recognizing that the error is gone. The integral action then changes the output inappropriately. In the case where dead time is the dominant dynamic element in the control loop, some type of dead time compensation is required to improve control loop performance. Although dead time is present to some degree in nearly all process control loops, the dynamic behavior of most loops is dominated by one or more capacity lags in series with the dead time. As an approximation for the case where there are multiple lags, the process can be characterized by an effective dead time (θ), an effective lag time constant (τ), and a steady-state gain (K P ). Dead time compensation should be considered whenever the ratio of dead time to lag time constant (θ/τ) exceeds 0.5. The Smith Predictor controller modifies the process variable signal to a conventional PID controller. The controller output provides the input to the process model as well as the actual process. The process model has two components, one with dead time and one without dead time. If there is a good match between the dynamics of the model and the process, the output of the model with dead time will cancel the output of the process. The process variable signal that remains for the controller will be the output of the model without dead time. This has the effect of mathematically eliminating the dead time from the control loop. The controller can then be tuned and the loop should perform as if the process had no dead time. The tuning of the Smith Predictor is based on the process characteristics (θ, τ, and K P ). The effective lag time constant (τ), in minutes, is entered into the PROCESS_LAG block as the TLAG softlist parameter. The effective dead time (θ) is entered in the PROCESS_DT block s DT softlist parameter. The steadystate gain (K P ), in units per units, is entered in the KP block. The PROCESS_SUB block calculates the difference between the process model without the dead time and the model with the dead time. The PROCESS_MUL block adds the scaling necessary to complete the model and then the _ADD block adds the to the output of model. The inputs and outputs of the SINGL_LP_SM_PR block include: = process variable = setpoint = output January

25 SINGL_LP_SM_PR CG39ARCAL TUNING Tuning involves first entering the lag time constant (τ), in minutes, in the PROCESS_LAG block s TLAG softlist parameter. Then the dead time value (θ) is entered in the PROCESS_DT block s DT softlist parameter, and then the steady-state gain (K P ), in units per units, is entered in the KP block. The controller is then tuned using conventional tuning methods ANALYSIS OF SMITH PREDICTOR CONTROLLER The Smith Predictor has the potential to provide perfect dead time compensation. However, its performance depends entirely on the accuracy of the model. Therefore, it is less robust than either the step and wait controller (see Section 13.0) or the complementary feedback controller (see Section 14.0). It should also be noted that the Smith Predictor provides better performance for setpoint changes than it does for load changes. On setpoint changes, the transient response of the internal model is in sync with that of the process since the source of the disturbance is the controller output. On load changes, the disturbance may be affecting the process up to one full dead time before the controller sees it. If the model response is out of sync with the process, it cannot do as effective a job of dead time compensation. For more information on the Smith Predictor, see Dead Time Compensation Using The Model 352 Single- Loop Controller, AD January 1999

26 SINGLE_LP_CF_SR 16.0 SINGLE_LP_CF_SR (Split-Range Coarse/Fine Single Loop) The derived block SINGL_LP_CF_SR is a split-range coarse/fine single loop controller. It is a modification of the SINGLE_LOOP Basic block with the controller output value converted into two output values, one output manipulating a coarse valve and the other manipulating a fine valve. A coarse/fine control strategy uses two final control elements (FCEs), one large and one small, connected for an additive effect on the process. The large FCE supports large changes in the manipulated variable but often lacks resolution. To improve resolution and increase turndown, a small FCE, which has better resolution, is used to trim the large FCE. Coarse/fine control is sometimes called big valve/little valve control. The split-range control strategy involves closing the coarse valve and manipulating only the fine valve when the flow demand is low. As flow demand increases, the fine valve is opened fully and remains open as the coarse valve is manipulated. This scheme features immediate response to demand changes and fine resolution for low flow demands. However, the resolution becomes coarse when the demand exceeds the capacity of the fine valve. This is a simple way to provide high turndown. The X_HOLD block holds the value of X, where: X = Max.Fine Flow Max. Fine Flow + Max. Coarse Flow The equations in cells N3 and N4 calculate the fine and coarse output values. The inputs and outputs of the SINGL_LP_CF_SR block include: = process variable _F _C = setpoint = controller output = output to the fine FCE = output to the coarse FCE For more information on split-range coarse/fine control, see Coarse/Fine Control Strategies Using The Model 352 Single-Loop Controller, AD January

27 SINGLE_LP_CF_FL 17.0 SINGLE_LP_CF_FL (Floating Coarse/Fine Single Loop) The derived block SINGL_LP_CF_FL is a floating coarse/fine single loop controller. It is a modification of the SINGLE_LOOP Basic block with the controller output value converted into two output values, one output that manipulates a coarse valve and one that manipulates a fine valve. A coarse/fine control strategy uses two final control elements (FCEs), one large and one small, that are connected for an additive effect on the process. The large FCE supports large changes in the manipulated variable, but often lacks resolution. To improve resolution and increase turndown, a small FCE, which has better resolution, is used to trim the large FCE. Coarse/fine control is sometimes called big valve/little valve control. The floating control strategy only adjusts the fine valve while its position is within fixed limits (e.g. between 25% and 75%) and adjusts the coarse valve only when required to keep the fine valve near midstroke. In this strategy, a controller block manipulates the fine valve. When the controller output exceeds the limits on the fine valve, a separate integral-only controller moves the coarse valve. The rate at which the coarse valve moves is relatively slow and is proportional to the difference between the fine valve signal and its exceeded limit. As the coarse valve moves, the fine flow controller responds by returning the fine valve to within its limits, at which time the coarse valve holds at its new position. The coarse flow controller must be tuned for slow response to ensure that the fine flow loop remains stable. This strategy responds quickly to small flow demand changes over the entire demand range by directly adjusting the fine valve. However, it responds slowly to a large demand change that saturates the fine valve. This is an easy way to get accurate flow control over a large flow range. The CONTROLLER block output manipulates the fine valve. The _LIMIT block determines if the output to the fine valve has exceeded its limits. The FLOAT_CONTROLLER block (an ID controller) will change its output (the coarse valve signal) whenever the fine valve output is limited. The inputs and outputs of the SINGL_LP_CF_FL block include: = process variable _C = setpoint = output to the fine FCE = output to the coarse FCE 17.1 TUNING The controller is tuned using conventional tuning methods. Be aware that the FLOAT_CONTROLLER must be given a large value of TI to ensure that the control loop remains stable. For more information on floating coarse/fine control, see Coarse/Fine Control Strategies Using The Model 352 Single-Loop Controller, AD January

28 SINGLE_LP_CF_CS 18.0 SINGLE_LP_CF_CS (Center-Seeking Coarse/Fine Single Loop) The derived block SINGL_LP_CF_CS is a center-seeking coarse/fine single loop controller. It is a modification of the SINGLE_LOOP Basic block with the controller output value converted into two output values, one output that manipulates a coarse valve and one that manipulates a fine valve. A coarse/fine control strategy uses two final control elements (FCEs), one large and one small, that are connected for an additive effect on the process. The large FCE supports large changes in the manipulated variable, but often lacks resolution. To improve resolution and increase turndown, a small FCE with better resolution is used to trim the large FCE. Coarse/fine control is sometimes called big valve/little valve control. The center-seeking control strategy involves manipulating only the fine valve while its position is within a certain trip point. When the trip point is reached, the coarse valve is adjusted until the fine valve is returned to 50%. In this strategy, a controller block manipulates the fine valve. A separate integral-only controller moves the coarse valve when the controller output exceeds the limits on the fine valve. The rate at which the coarse valve moves is relatively slow and is proportional to the difference between the fine valve signal and 50%. As the coarse valve moves, the fine flow controller responds by returning the fine valve to 50%, at which time the coarse valve holds at its new position. The coarse flow controller must be tuned for slow response to ensure that the fine flow loop remains stable. This strategy responds quickly to small flow demand changes over the entire demand range by directly adjusting the fine valve. However, it responds slowly to a large demand change that saturates the fine valve. The advantage of center-seeking control is that it returns the fine valve to where it can respond alone to large changes in either direction. The CONTROLLER block output manipulates the fine valve. The _COMPARATOR block determines if the output to the fine valve has reached a trip point (e.g. 25% or 75%, which is represented by a 25% deviation from 50%) and if the fine valve has returned to 50% via the DBAND softlist setting of 25%. The FLOAT_CONTROLLER block (an ID controller) will change its output (the coarse valve signal) whenever the fine valve output has reached a trip point and has yet to return to 50%. The inputs and outputs of the SINGL_LP_CF_CS block include: = process variable _C = setpoint = output to the fine FCE = output to the coarse FCE January

29 SINGL_LP_CF_CS CG39ARCAL TUNING The controller is tuned using conventional tuning methods. Be aware that the FLOAT_CONTROLLER must be given a large value of TI to ensure that the control loop remains stable. For more information on center-seeking coarse/fine control, see Coarse/Fine Control Strategies Using The Model 352 Single-Loop Controller, AD January 1999

30 BAL_2FE 19.0 BAL_2FE (Balance of Two Final Control Elements) The derived block BAL_2FE balances a load between two parallel final control elements, each of which has an adjustable bias value and the ability to switch to a manual output. The block uses an integral-only controller to adjust the load to each BIAS_FE block (see Section 21.0) so that the desired total output is achieved. The POSITION_CNTRLR block is the integral-only position controller (ID without derivative). The integral time of the controller can be set at a very small value (0.01 min) without instability since there is no process dynamics in the position loop. The [T_] output is calculated each scan from the BIAS_FE block outputs of that scan. The BIAS_FE block outputs are first scaled to the correct percentage of the overall capacity then the correct feedback for the position controller is calculated based on the number of final control elements that are running and in auto mode. If the final control elements are not of equal capacity, the SCALER block Soft List values should be changed to reflect the correct proportion of the total output that each element contributes. The T_ output (total output) is provided for the feedback of an upstream controller (e.g. the PRIMARY Basic block). The output is provided to indicate that at least one of the final control elements is in auto mode. If this condition is not met, the upstream controller should track. The inputs and outputs of the BAL_2FE block include: PROOF1 PROOF2 = setpoint (desired total output) = running status of final control element 1 = running status of final control element 2 1 = output to final control element 1 2 = output to final control element 2 T_ = scaled total of the individual outputs (for feedback of an upstream controller) = auto status (indicates one of the final control elements is in auto) January

31 BAL_3FE 20.0 BAL_3FE (Balance of Three Final Control Elements) The derived block BAL_3FE balances a load between three parallel final control elements, each of which has an adjustable bias value and the ability to switch to a manual output. The block uses an integral-only controller to adjust the load to each BIAS_FE block (see Section 21.0) so that the desired total output is achieved. The POSITION_CNTRLR block is the integral-only position controller (ID without derivative). The integral time can be set at a very small value (0.01 min) without instability since there is no process dynamics in the position loop. The [T_] output is calculated each scan from the BIAS_FE block outputs of that scan. The BIAS_FE block outputs are first scaled to the correct percentage of the overall capacity then the correct feedback for the position controller is calculated based on the number of final control elements that are running and in auto mode. If the final control elements are not of equal capacity, the SCALER block Soft List values should be changed to reflect the correct proportion of the total output that each element contributes. The T_ output (total output) is provided for the feedback of an upstream controller (e.g. the PRIMARY Basic block). The output is provided to indicate that at least one of the final control elements is in auto mode. If this condition is not met, the upstream controller should track. The inputs and outputs of the BAL_3FE block include: PROOF1 PROOF2 PROOF3 = setpoint (desired total output) = running status of final control element 1 = running status of final control element 2 = running status of final control element 3 1 = output to final control element 1 2 = output to final control element 2 3 = output to final control element 3 T_ = scaled total of the individual outputs (for feedback of an upstream controller) = auto status (indicates one of the final control elements is in auto) January

32 BIAS_FE 21.0 BIAS_FE (Bias of a Final Control Element) The derived block BIAS_FE performs the function of a final control element auto/manual biasing station. It works in conjunction with other BIAS_FE blocks inside a BAL_2FE block (see Section 19.0) or BAL_3FE block (see Section 20.0). The input load signal can be biased to become the output, or a manual output can be selected. The bias value is comprised of two parts, the absolute value of the bias and the action (positive or negative) of the bias. This configuration prevents a large bias change from being entered if the sign is accidentally left out. The POSITIVE_BIAS block holds the action of the bias: TRUE is positive and FALSE is negative. The SETPOINT block holds the absolute value of the bias. Upon a change in bias action, the setpoint (bias value) is set to 0% via the TRIG blocks and the EXTRNL input to the SETPOINT block. The bias is calculated and added to the load input () at the BIAS_ADD block. The M2A_BALANCE block (a PD block) is used to prevent a bump in the output when switching from manual mode to auto mode. The PD block has a lag built in that is enabled on the switch to auto mode. The time constant of this lag is the MR_TLAG Soft List parameter. If the final control element is not running (PROOF input is FALSE), the circuit rejects to manual mode and the output is set to 0%. The inputs and outputs of the BIAS_FE block include: PROOF = process variable (unbiased load signal) = running status of final control element BIAS = output to final control element = bias value including the sign (positive or negative) January