Control Loop Tuning

Hello, 

I have a puzzling question or concern. I am still a bit new to Automation, but learning. 

Often operators at my plant will show me a control loop (flow, pressure, etc) and state the the valve is swinging (oscillating) all over the place. Then there is a request for me to tune the loop. 

One of my biggest concern is that, "what if there is some other instrument, I/O communication, valve, positioner, etc malfunction that causing the overshoot or instability in the loop". I don't want to just start tuning a loop without proper investigation of the entire loop to make sure maybe valve is not funcitoning correctly. 

Is there anything I can read related to proper diagnoistics of a valve assembly as it relates to tuning a loop?

Is there a program in DeltaV I can use to monitor the loop performance to see how the loop performs for all process conditions?

Like now, I have a loop with PV Scale of 0 - 3000 pph, loop is in manual, the PID out is 10% (and valve is showing 10% open), and the PV is 1600 pph. It doesn't make sense to me how the valve can be only 10% open while operating at 1600pph which is the mid range of the scale. maybe I am confusing my self. 

Thanks for the help.

16 Replies

  • HI Darrell and welcome to the 365 Community. Happy to see Yousef and others providing such great responses. I will also share your question in our Valves group and see if they can offer any advice. In the interim, please reach out to me with any suggestions or thoughts on how we can improve your experience in 365.

    Look forward to learning more from you during your journey towards Automation Expert!

    Best Regards,

    Rachelle McWright: Business Development Manager, Dynamic Simulation: U.S. Gulf Coast

  • The PV has a lot to do with pipe size pressure and type of valve.  With high upstream pressure with say a butterfly valve even with the valve only partially open you could get a significant flow.  Now change the butterfly to a V-ball and I wouldn't expect to see much flow in the same system.  You can also check historical data do you have the same flows and equipment CV's for the same / similar conditions or has something significantly changed.  just my 2 cents.  Most valves do no exhibit linear flow to CV characteristics.

    best analogy I can think of is put your finger over the end of a garden hose and only leave 10% exposed lots of water comes out at high velocity.  Then take your finger off and still a lot of water comes out (more than when you were covering the hose but still alot) at a low velocity but there is still a high volume flow rate with both cases.

  • Use DeltaV Tune with Insight. You can find that application under DeltaV - Advanced Control

    When tuning a loop, it is best to check first the condition of each element. Is the valve gate having stiction? Does the valve opens and close with respect to the signal being provided? Are there any deposited materials in the pipes? Does the transmitter provide the correct output current for a given valve position?

    Once you have cleared all these checks and still the loop does not behave the way it should be, or you have already reached the limitations/constraints of each elements, then that's the time that we tune it using the tuning parameters (i.e. gain, reset time, derivative time, dead time, lead/lag etc).

    As for the last item, the OUT of the PID does not correspond to the value of PV with respect to it's operating range. It tells you how the PID would control the output element with respect to the error (Error = SP - PV). Now, there are certain factors that would affect how the OUT is calculated. It would depend on the algorithm or tuning type you are using (ex: P-only, PI, PID).

    I would recommend reading some articles on PID tuning which can be found on Books Online.

  • In reply to David Asuncion:

    Darrell,

    The best way to evaluate a loop is to put it in Manual.  Unless the process is a runaway or integrator (Non-Self_Regulating) it should steady out.  If it doesn't, then you have a problem with the positioner or actuator.  Once it is steady, you make make a small manual change and see if it responds.  Depending on the valve, you may not see a change until you have made 0.5, 1.0, or larger % change in output.  There may also be a delay in how long after the controller output is changed before you see any movement at all.  You can try moving the valve up and down the range (Operations permitting) and change directions to understand how well the valve performs.  You can also calculate the valve gain, which may not be linear (hence the reason for making step changes over the normal range of operation.

    There are various methods for tuning loops.  Some are done closed loop like the DeltaV On-Demand tuner in Insight.  Some are done open loop as I described above.  Emerson has a group of consultants that provide loop analysis and tuning around the world and across all industries.  They are collectively known as Control Performance consultants and many of them will be at Emerson Exchange.  Emerson's Educational Services also offers a course called Modern Loop Tuning that is an excellent way to learn about tuning loops.  A short version of this course is usually offered as a short course at Emerson Exchange.

    It is pretty easy to identify or rule out instrument and valve problems.  If the loop elements are working and the process is not well controlled, it may be tuning.  But remember that there are things a PID loop cannot do very well.  One is to respond quickly without oscillation when there is significant deadtime in the process.  Another is to respond effectively when there are multiple loops interacting.  These can be managed, but it takes some understanding.  Process Insight is a great first step.  It can learn process dynamics while the process is running and it can allow you to tune on demand.  It also can track loop performance and provide an indication or report on which loops are the most under-performing.  It fits the 80-20 rule nicely.

    BTW, it is purely coincidental if 0 and 100% Valve position correspond to 0 and 100% of transmitter span.  That may be the design and specification goal, but there are many factors which will affect it.

  • This website has a wealth of information about controls control valves. http://blog.opticontrols.com/.

    Also if the pid controller is cascaded (ie. output is anything other than 0-100%) make sure the ARW_HI_LIM an ARW_LO_LIM are configured correctly.

  • In reply to Lou Heavner:

    Hola Darrel, the 10% valve opening giving you a 50% of your operating range suggest me you could have a oversized control valve. Working in the low and high region of the valve characteristics will give the controller a hard time. See an extract from the Emerson CONTROL VALVE HANDBOOK. ( www.controlglobal.com/.../wp_06_059_fisher_valvehbook.pdf )

    "...oversized valves hurt process variability is that an oversized valve is likely to operate more frequently at lower valve openings where seal friction can be greater, particularly in rotary valves. Because an oversized valve produces a disproportionately large flow change for a given increment of valve travel, this phenomenon can greatly exaggerate the process variability associated with dead band due to friction."

    I am assuming your transmiter is properly calibrated and reading the right values.

    Make sure your positioner has no integral action enabled. If you still suspect of interactions try to identify those control loops that could be affecting, diferentiate which ones need to be fast and which ones need to be slow loops; if tuning is not considering this you will end up with loops fighting each other. Lambda tuning could be a good way to solve interactions. You can find more info here: www.emersonprocessxperts.com/.../lambda-tuning

    Hope this helps,

    Héctor

  • In reply to HECTOR H. TORRES:

    Thank you all for responding.

    Currently the loop is Manual. When the loop is placed in Auto, it instantly maintains the setpoint (the same OUT % as in Manual mode). If a step change of about 300 pph, the PV starts oscillating. I need to investigate what the PID  OUT does.

    I will check the valve assembly to determine if there is a malfunction. Is there a good reference for checking the valve assembly to confirm there are no issues? I do have the Fisher Control Valve book. Maybe a good I/E tech book.

    Can an oversize valve cause oscillating or control loop instability?

    Thank you

  • Darrell,
     
    If you don’t mind me saying so, it doesn’t appear to me that you are “a bit new to automation” at all!  In my experience, control valve and positioner faults and limitations cause more problems than just about everything else combined, so you are absolutely correct to check the valve first.  If your valve positioner has HART capability, it should be feeding back the actual valve position (as determined by the positioner feedback arm) to the DeltaV system, where this should appear as a parameter of the AO (analog output) function block.  If so, you can do some tests to check for valve/positioner problems (stiction, hysteresis, slow slew rate etc.).  Trend both the commanded valve position (analog output) and the actual position (positioner feedback) parameters and make some manual output changes.  Then put the loop in Auto and make it work a bit (e.g., change setpoint or load).  Look for dynamic discrepancies between output and feedback.  You cannot correct valve/positioner problems by tuning – remember the first rule of computing: garbage in, garbage out...
     
    On your other point, I don’t think your expectation that the PV should be at the same percentage of range as the valve output is correct.  (It would be great if this always were true!)  Many things can affect what valve position is required to achieve a given PV (valve characteristic, process load etc.).  You don’t say what type of loop it is, but it rather sounds like a flow controller.  Nor do you say what the approximate period of the oscillations is.  If it is a flow, I’d expect a relatively low gain (perhaps 0.5 – 1.0, but this depends on the valve), a short integral time (2 – 5 seconds) and no derivative.  Have you checked for ‘measurement noise’?  If the PV is noisy, you should apply a PV filter – but not too much if it is a flow controller.  Look at the raw PV on the trend system with a 1-sec refresh rate.  If it’s noisy at all, apply a light filter (try 1 second; I wouldn’t use much more than 2 seconds on a flow control loop).
     
    Best regards,
     
    Neil Brown
     
    From: darrell
    Sent: Wednesday, July 02, 2014 4:11 PM
    Subject: [EE365 DeltaV Track] Control Loop Tuning
     

    Hello,

    I have a puzzling question or concern. I am still a bit new to Automation, but learning.

    Often operators at my plant will show me a control loop (flow, pressure, etc) and state the the valve is swinging (oscillating) all over the place. Then there is a request for me to tune the loop.

    One of my biggest concern is that, "what if there is some other instrument, I/O communication, valve, positioner, etc malfunction that causing the overshoot or instability in the loop". I don't want to just start tuning a loop without proper investigation of the entire loop to make sure maybe valve is not funcitoning correctly.

    Is there anything I can read related to proper diagnoistics of a valve assembly as it relates to tuning a loop?

    Is there a program in DeltaV I can use to monitor the loop performance to see how the loop performs for all process conditions?

    Like now, I have a loop with PV Scale of 0 - 3000 pph, loop is in manual, the PID out is 10% (and valve is showing 10% open), and the PV is 1600 pph. It doesn't make sense to me how the valve can be only 10% open while operating at 1600pph which is the mid range of the scale. maybe I am confusing my self.

    Thanks for the help.

  • In reply to neilrbrown:

    Exceptional advice, Neil Brown. Thanks so much for your thoughtful response!

    Best Regards,

    Rachelle McWright: Business Development Manager, Dynamic Simulation: U.S. Gulf Coast

  • Hi Darrell,

    As loop tuning rarely "goes bad", what I tell my team is that when asked to tune a loop there are some things they should consider.

    Did this loop ever work well or has it always had these problems?  

    * If the loop has never worked well, or has been recently commissioned, tuning is likely the appropriate response, although there may be design issues causing the instability.

    If the loop used to work well all the time, what has changed?  

    * If something has worn and needs maintenance, fixing the problem is preferable over tuning, although circumstances may require temporary changes in tuning until maintenance can be performed.

    * If something has been replaced and has affected response, re-tuning the loop for the new equipment is likely necessary although some investigation as to whether the replacement is appropriate may be in order.

    * If the process itself has changed, it's important to understand what has changed and why.  If the process dynamics have changed enough to affect loop performance then tuning is appropriate.

    * Has the loop tuning been changed previously and that is causing the problem?  Poor change management or loss of online changes through a system cold restart may require re-tuning or returning to previous parameters.

    Is the loop oscillating or unstable some of the time but works well the rest of the time?

    * Check for design issues - improper valve sizing for example (as already mentioned) may result in instability when a valve operates at or near its limits.

    * Review process and controls - process gain may vary enough that a single set of tuning parameters is not adequate, or a measurement or response needs to be characterized to compensate for non-linearity.

    A few people have mentioned DeltaV Insight for loop tuning.  You can also use DeltaV Insight for monitoring your control loops for performance and to identify common problems - cycling and instability, operating at limits, valve stiction, etc.  Insight is an optional package which may require purchasing a license.  See the product data sheet at: www2.emersonprocess.com/.../PDS_DeltaV_Insight.pdf

    Cheers,

    Kevin

  • In reply to Lou Heavner:

    Thank you LouHeavner for the response and advice.

    What means "It fits the 80-20 rule nicely"

    Thank you.

  • In reply to DCS Newbie:

    From Wikipedia:

    The Pareto principle (also known as the 80–20 rule, the law of the vital few, and the principle of factor sparsity) states that, for many events, roughly 80% of the effects come from 20% of the causes. Management consultant Joseph M. Juran suggested the principle and named it after Italian economist Vilfredo Pareto, who observed in 1906 that 80% of the land in Italy was owned by 20% of the population; Pareto developed the principle by observing that 20% of the pea pods in his garden contained 80% of the peas.

    It is a common rule of thumb in business; e.g., "80% of your sales come from 20% of your clients". Mathematically, the 80–20 rule is roughly followed by a power law distribution (also known as a Pareto distribution) for a particular set of parameters, and many natural phenomena have been shown empirically to exhibit such a distribution.

    In this instance though, I took liberties with the expression and meant that DeltaV Process Insight can help you with 80% of your loops.  For the other 20%, you may need more tools and techniques.  Since Insight is embedded in DeltaV and pretty easy to use, it should be significantly less cost and effort than other solutions.

  • In reply to Lou Heavner:

    I don't know what you think... but would you also say that the Pareto principle can apply to the workforce in that 80% of the people will let the vital 20% do all the work?  I know that at some of the jobs I've had that sometimes feels like the case.

    Natalie Namisnyk

  • In reply to HECTOR H. TORRES:

    Thank you Hector,

    To check for valve oversize, I would like to check the valve sizing calculations (or valve specification sheet) and check what Cv and flow values were used for calculation. Then confirm these values on the actual valve faceplate.

  • In reply to Lou Heavner:

    Thank you LouHeavner,

    I agree, Insight is something I need to read about and start using. Many loops I come in contact with have tuning issues.