Recent Control & Safety Systems Discussions
Similar Posts
About self-regulating and integrating questions
Recall: [EXTERNAL] RE: [EE365 DeltaV Forum] About self-regulating and integrating questions: Suggested Answer
Commissioning PI Control – Integrating Process
about ratio module
Question about OPC Licensing
Share
  • Not Answered

About integrating process

  and Forum's experts

About Integrating Process response's Tuning， I met some problems. There is a article about Integrating Process response by James, Based on this articles:  "Kp = (final slope – initial slope)/ Δ%OUT",  I calculate My KP  as follows:

First time, I gave a 20% Step Out, then measure PV value's slope, First Slope=1.994, Second, I gave a 10% Step Out and test it's slope again Second Slope=0.922, I think the Δ%OUT=10%, Then I calculate my Kp=(0.922-1.994)/10%= -10.72 , Based on this test result I got the Integral time and Controller gain, but these parameters cannot let my PID reach stead state. I think Root Cause is my Kp value. I noticed James's Kp= 0.000216, it's very small, but my Kp value is big, I think my calculation exist problems.

In addition, I notice James's  APD=30 Percent, How to select the APD value, it's value=Dead time?

Another question: How to select MLD, I notice the article using 40%, this value is a experienced Parameter?

I will appreciate if you can give me a help.

Best Regards

Jimmy Lee

13 Replies

  • James Beall
    Jimmy,-
    I'll provide more explanation later but here is a start. First, check your math, ((0.922-1.994)/10%= -0.1072, not -10.72. Also, slopes have units of per time. In fact, this slop should be in %PV/seconds. I like to use seconds because DeltaV Reset time has units of seconds. Also, the time units of dead time and Lambda must have the same units of time as the process gain (seconds) in this example. Please recheck your calculations and I will provide more on "Allowable Process Deviation (ADP)" and "Maximum Load Disturbance (MLD)" later.
    Good luck!
    James
  • Jimmy Lee
    Many Thanks for your Kindness, James.
     
    Now I descripted my test Procedure.
     
    I used a simulate Function Block(Not DeltaV Simulate Function Block) and try to use your example configuration parameter:
     
    Below Snapshot is my Simulate Function setup
     
     
    My Process Deadtime=30 seconds
    Process Time Constant=6900(I used Your example  Lamda value), Is it correct? If it’s not correct, how to get Process Time Constant?
    My PID SetPoint=0.3 KPa, PV Maximum=1 KPa
    My PID direction is Direct Action.
    My PV Value PT1001 is very slowly increasing, it seems correct, but Controller Output only have a little change from 100% to 99%
     
     
    By the way, If I used same tuning parameters, PID Direction is Reverse Action, the PV Value is very fast exceed my setpoint and always increasing and cannot reach stead state. I am not sure what’s wrong with my setup.
     
    Best Regards
     
    Jimmy
     
  • James Beall

    In reply to Jimmy Lee:

    Jimmy,
    The integrator model should not have a time constant (an integrator plus lag model does have a time constant from the process). The large (3600 seconds) time constant in your model will significantly impact the response. I am not familiar with the Simulation block that you are using but it seems odd that it has parameter "FOPDT" (usually meaning "first order plus deadtime self regulating) with the parameter value "Integrator". Also, the integrating gain must have "per time" units. What time units are in this block gain? Can you provide more details on this simulation block? An integrating process, and thus a simulation of the integrating process, integrates the difference between the "process demand" (which determines how much controller output is required to stop the PV from rising or falling). It looks like your SUB_REAL block may be taking the difference of 100 and your controller output, and thus "100" is the process demand. This means the controller output has to be 100 to stop the PV but it can only move the PV in one direction. I believe your "100" on this block should be something like 50. Thus, with the controller OUT at 50, the PV does not rise or fall.

    You could test the functionality of the Simulation block by setting the time constant to 0, the process gain to 0.01, perform a step test, and analyzing the step test trends to determine the actual integrating gain of the Simulate block.
    James
  • James Beall

    In reply to James Beall:

    I noticed your PV scale is 0-1 KPA. Note that all of the calculations for process gain and tuning MUST use %PV and %OUT. And, your integrating simulation should use %PV/second/%OUT to make the simulation process gain match your calculated process gain in units of %PV/second/%OUT.
    James
  • Jimmy Lee
    James,
     
    Many thanks for your help.
    Sorry response your late because I just back from business trip.
     
    1. You reminded me,  because my PID PV value’s range is actual Engineering Unit rather than 0-100, my simulation Block Process gain is 0.1, it generate PV is Engineering Unit, not PV%.
    By the way, You mentioned “Note that all of the calculations for process gain and tuning MUST use %PV and %OUT”, Do you mean I need to scale my Engineering Unit PV value into 0-100? For example: My transmitter Limit from 0-2MPa, I need to scale my PV value into 0-100?， so My PID feedback PV value=25*Engineering Unit’s value, Setpoint=25*real Setpoint? I noticed your simulation block seems using the PV’s value from 0-100. Please see one of my PID setup
     
     
     
    1. The reason I used SUB_REAL block, because If my PID direction is reverse acting, my Simulation block fit my PID, but if My PID direction is direct acting, my Simulation block can not match My PID, I have to add a Sub blocks.
     
    1. for the Integrating process, how to measure dead time? I can measure initial slope and final slope with difference step out, I mean if I am in customer field site to measure a real process response.
    2. Can you provide more on "Allowable Process Deviation (ADP)" and "Maximum Load Disturbance (MLD)" later.
     
    Next days I will continue to test my Integrating and self-regulating process’s tuning,  If I met some issues, I will let you know.
     
    Jimmy
     
  • Jimmy Lee
    Today, I test the Self-Regulating Process Response again, First time, I used actual Engineering Unit as my PV Range , for example, 0-4MPa, Second time, I scaled the 0-4MPa into PV 0-100, and corresponding Setpoint multiply 25, Both test results are perfect. And I understand why you said Process Gain should be  in 0.5 to 2, because I used PV range is actual Engineering Units in the past, so this Process Gain is very small.
     
    By the Way, Yesterday, I made a mistake about My Transmitter Range, It should be 0-4MPa, so My setpoint should be 25*SetPoint based on Engineering Units.
     
    Below is my test calculation snapshot, I also attached the excel file to you.
    .
    PV Range
    Process Gain(Kp)
    Tau98
    （Second）
    Tau
    （Second）
    Td
    (Dead Time    ）
    PID Step Out
    λ
    Controller Gain(Kc)
    Integral Reset
    （Repeats/Minutes)
    PV:0-4MPa; SetPoint: 1.6
    0.04
    77
    19.25
    9
    10%
    57.75
    7.209737828
    3.116883117
    0.04
    77
    19.25
    9
    10%
    38.5
    10.13157895
    3.116883117
    0.04
    77
    19.25
    9
    10%
    19.25
    17.03539823
    3.116883117
    0.04
    77
    19.25
    9
    10%
    9.625
    25.83892617
    3.116883117
     
     
     
     
     
     
     
     
     
    PV:0-100; SetPoint: 40
    1
    78
    19.5
    8
    10%
    58.5
    0.293233083
    3.076923077
    1
    78
    19.5
    8
    10%
    39
    0.414893617
    3.076923077
    1
    78
    19.5
    8
    10%
    19.5
    0.709090909
    3.076923077
    1
    78
    19.5
    8
    10%
    9.75
    1.098591549
    3.076923077
     
     
    Best Regards
     
    Jimmy
     

    Lambda_Tuning_Parameter.xlsx

  • James Beall

    In reply to Jimmy Lee:

    Jimmy,
    It is perfectly fine to have PV and/or OUT scale that are not 0-100%. You just have to convert the EU PV or OUT to % of its own scale for all of the process gain and tuning calculations. So, if the PV or OUT is not 0-100%, you must convert as follows:

    %PV= PVEU * (100%/ PVEU Span).

    %OUT= OUTEU * (100%/ OUTEU Span).

    Back to your integrating process simulation, can you provide any documentation on the simulation you are using? Next week I can send DeltaV configuration for an integrating process simulator.
    James
  • Jimmy Lee
    James，
     
    Please check  page 190 in attached, there is a Simulate Block instruction’s description,  Page 112 is ADV_PID instruction’s description.
     
    My Simulate Block is very simple, It can be configuration as Self-Regulating or Integrating Model.
     
    If I configurate Process Gain=0.1 under “Integrating Model”, Please look at below snapshot, I used a data record capture the chart data, I used the formula that you provided “ Kp = (final slope – initial slope)/ Δ%OUT”, This calculation result just right  is 0.1.
     
    In addition, I want to know if the Valve initial position should not be  0% or 100%  for a Integrating process, true or false.
     
    By the way, In the past, I was in GE PLC department, about four years ago, Our department was sold to Emerson, When We joined the Emerson, we belonged to Emerson PSS team, now we were re-org to FLMC team
     
     
    Best Regards
     
    Jimmy
     
  • James Beall

    In reply to Jimmy Lee:

    Hi Jimmy,
    I don't see an attached manual on the block. What are its units of integration (e.g. seconds)?

    "The controller output (aka valve position) to stop the PV from rising or falling is set by the "load" or "throughput" through the system. Think about a tank level where the controller adjusts the inlet valve to hold the level. The inflow must match the outflow to hold the level. So, in this case, the outflow is the "load". In your simulation, the "IN1" on the SUB REAL block represents the load on your simulation. When the controller output = IN1, then the SUB REAL output is 0 and thus the integrator block's output stops moving (integrating a 0 input). Note that if you controller output is 0-100%, the output of the integrator block is %PV but if you don't scale it to the PV EUSCALE (0-100% input = PV EU Scale), it will still work but the integrating gain you calculate will be different than the integrating gain you put into the simulator block. For the integrator simulation, I usually scale the controller output to 0-100% (if not already) and then convert the output of the integrator block from % to PVEUScale.
    Note that integrating gains are typically very small, I would use no larger that 0.01%PV/second/%OUT in your simulation.

    So, MLD = maximum load disturbance. Think about the tank level example I gave above. MLD is the maximum step (or near step) move that you expect of the Load THAT YOU WANT THE CONTROLLER TO KEEP THE PV WITHIN THE ALLOWABLE DEVIATION LIMIT (ADL) WITHOUT OPERATOR INTERVENTION. Suppose you review the history data of the loop, talk to the operator and find out that the MLD is about 20% as seen by the controller out that counteracts the load change. Thus, MLD=20%OUT. For this MLD, you choose the maximum ALLOWABLE DEVIATION LIMIT in %PV units, say 15%PV. Then you can use the equation I gave to calculate the required maximum Lambda (i.e. how aggressive does the controller tuning need to be) to handle the MLD before the %PV has deviated the ADL. Note that the PV scale doesn't have to be 0-100% but you have to convert you EU's to percent for these calculations. For example, for the 0-4 MPa scale, if you only want ALDeu=0.2 MPA, ALD%=0.2 MPa*(100%PV/4 MPa) = 5%PV.

    You might want to initially set the PV scale to 0-100% and get proficient at the calculations and results. Then switch the PV scale back to 0-4 MPa, scale the output of the integrator from 0-100% to 0-4 MPa and you should get the same Kp and tuning.

    Keep at it! You'll get it!

    James
  • Jimmy Lee
    I did attach file to you, but I don’t know why no upload, can you give me a mailbox, I can send it to your mailbox .
     
    “What are its units of integration (e.g. seconds)?” Answer: it’s second.
     
     
    Best Regards
     
    Jimmy
     
  • Jimmy Lee
    Maybe my manual’s size is bigger(3M), the forum have the file’s size limitation, now I only pick up this block description and attach to you.
     
    Best Regards
     
    Jimmy
     

    Simulate.pdf

  • James Beall

    In reply to Jimmy Lee:

    Got the smaller file!! Thanks!
    James
  • JoshC

    In reply to James Beall:

    I'm interested in the DeltaV configuration for an integrating process simulator if you're willing to share.

This is the official online community site of the Emerson Global Users Exchange, a forum for the free exchange of non-proprietary information among the global user community of all Emerson Automation Solution's products and services. Our goal is to improve the efficiency and use of automation systems and solutions employed at members’ facilities by sharing our knowledge, experiences, and application information.

User Groups | World Areas | Community Guidelines | Legal Information | Privacy Policy | Contact Community Manager

Website translation provided by Website Translator

© 2015-2022 Emerson Global Users Exchange. All rights reserved.