# Control Loop Performance

J

#### Jim Ryan

I've read a few articles on control loop performance. We have over 2500 control loops in our plant and interested in monitoring atleast 800 of them online.

Does anyone have any experience with any of the commercial packages out there ?

JR

H

#### Hans Eder

I am glad you bring up the subject: Proper monitoring the dynamic performance of key process variables - controlled automatically or not - is a key success factor and the basis to quantify the success of any effort, may it be the re-tuning of a controller or replacing it by another one or an entirely different scheme. Unfortunately, even today the DCS systems do not come already equipped with it - hard to understand.

When I was Advanced Control Manager, the Performance Measurement System (PMS) was a key instrument for me, also for work planning, e.g. answering questions like: do we need to improve some of the existing controls or can we go on developing new ones, and also to document and to advertise our success - not qualitatively ("the controller runs now much better") but in engineering units and in hard $figures: Doing the proper measurements it was then easy to claim without dispute that we have reduced the variance by x% and this has brought a certain increase in product yield etc. We have a PMS system designed that can be quickly implemented in the DCS by your own staff. Hans H. Eder ACT [email protected] and [email protected] www.act-control.com J #### Jim Ryan Thanks for the post. I agree with your comments. I will look at the website. Do you know of any chemical or refining companies using these types of tools? Has anyone published any case studies ? I've surfed the Honeywell (Loop Scout), Invensis (Loop Analyst), and Matrikon (Process Doctor) sites. Have you used or do you know about any of these tools? I would appreciate comments from anyone with first hand experience. regards, JR J #### John G. Boland Hi, Jim, I had the same problem during a decade of DCS installation and expansion. Tons of data. So much that it became impractical to determine "which loops to optimize today". Most were just fine, but still had to be checked. Eventually the guys just fixed broken loops and left the rest. VisiBit software addresses this problem. Please contact me if you are interested by the website. Best, John G. Boland, president VisiBit Corporation www.visibit.com One Parker Square Suite 408 2525 Kell Boulevard Wichita Falls, Texas 76308 940.322.9922 940.723.1478 fax L #### Lane Desborough Hi John, I'm the guy who co-authored one of the first academic papers on control loop performance monitoring with Dr. Tom Harris. Harris, T. J. Desborough, L., "Performance assessment measures for univariate feedback control", The Canadian Journal of Chemical Eng., Vol. 70 (1992), pp. 1186-1197. If you're interested in some more practical reading material, check out last fall's ISA Intech Article "Control System Reliability: Process Out of Control" at "http://www.isa.org/isaolp/journals/pdf/intech/20010852.pdf":http://www.isa.org/isaolp/journals/pdf/intech/20010852.pdf I've been working for Honeywell the last seven years and am the manager of Loop Scout, Honeywell's loop monitoring service. Please feel free to give me a call to discuss your particular needs. I can put you in touch with some of our customers who can discuss economic benefit. To use Loop Scout you need an OPC server for your non-Honeywell control system or a Honeywell DCS such as TPS, Alcont, or PlantScape. Regards, Lane Desborough [email protected] 805 777 8324 M #### Michael Wroe I am in charge of leading, managing, training, etc. of our loop performance program for my company. We are a global chemical company in the inorganic, petrochemical, and specialty chemical industries. We employ a software/hardware tool we purchased to perform some of the loop performance monitoring/auditing. However, due to its cost, it is a limited item. We have since developed some tools internally so that multiple people may analyze loop performance. The monumental point in the program was when our company went live with a global plant historian. This historian allows every employee see process data. Placing data in the hands of operations personnel (such as operators) and training them on how to identify issues with control performance adds a large number of people to the team. Although the operators are unable to improve the control performance, they are involved in the improvement process and generally want to help as much as possible to identify issues. In addition, the historian package has as an option a control monitor program that monitors the control performance of loops. This program is very time consuming to set up since one has to pick weight functions to rank control performance issues. However, once it is set up, it is useful for a process control group to use to identify the top ten loops to work on during any one time. This has been successfully employed at a few of our sites. One key thought you should keep in mind when attempting to find a starting point is that there are usually a limited number of key loops in a plant. Identifying the key loops and starting there may help progress you forward. My experience shows me that out of 2500+ loops, usually only 20% are actually key loops. Using a Manufacturing To Target program can aid in identifying your key loops as they relate to quality and production rates. G #### gilliland Jim You might also look at the ControlMonitor package at "www.ControlArtsInc.com":http://www.ControlArtsInc.com . Most packages check control loop performance by comparing your controller to a minimum variance controller. Trouble is, you can't ever install a minimum variance controller, so the results can be sub-optimal. The ControlMonitor package compares your controller to an "optimal" DMC-type controller or an "optimal" PID controller. The results are more sensitive and indicative of things you can change. There's also a methodolgy for evaluation level controllers - here the objective is to let the measurement swing so the controlled variable variance is minimized. D #### David B. Leach I suggest that you check out ExperTune's new control loop performance monitoring product, called Plant Triage ( "http://www.expertune.com/PlantTriage.html":http://www.expertune.com/PlantTriage.html ). I was formerly a control systems engineer and advanced control technology manager for a large industrial gases and chemicals co. for the last 22 years, and recently took the "early early" retirement pkg. so I am on my own now. I am in the process of becoming an independent consultant. I have worked with ExperTune as one of their most active customers over the last 15 years, and was an alpha and beta tester for many of their new product releases. I was able to work with ExperTune to set the design criteria for Plant Triage in the conceptual design phase, and it has many of my ideas included, spanning 32 years of industrial practice working for several large chemicals and gases companies. I evaluated "on paper" other competitive products such as Matrikon's Process Doc, which is a fine product, however as far as I know it does not have an imbedded PID loop tuner that has the same capabilities as the ExperTune Advanced Control Tuner, which is included in Plant Triage. Also Plant Triage includes a full featured, relatively user-friendly OPC-compliant historian. Note that both Matrikon's Process Doc and ExperTune's Plant Triage are fairly expensive software products (i.e., >$35K.) Also, as far as I know they are both sold with consulting services in order to set up the software in the plant and train users on how to properly use it, which is an essential component as far as I'm concerned. I've found that without training, even tuning-only pkgs such as ExperTune are generally not used to their full advantage to optimize plant control performance.

David B. Leach
leachdb @erols.com

G

#### Gino

Hi David,

We are doing an evaluation in this area on a number of the products. ExpertTune makes a great tuning package no doubt. Not so sure about Triage though ? I've not used it yet but understand that you have to implement a separate historical database to use it. At our facilities, we reluctant to support another database in the plant. We have PI, and IP/21 data historians across multiple facilities and would like to connect to them all from a single server. Data compression and scan rates will be adjusted accordingly of course. We would like to be able to connect to them all simultaneously from a single server with clients in remote locations. Our plants are fairly small with approximately only 20-30 loops in each plant worth monitoring.

We are also wanting to monitor our DMC controllers. We are DMCplus with approximately 1-2 controllers per plant. Again without APC engineers at each site, I want to leverage my skilled control engineers across multiple facilities. ProcessDoctor is now able to analyse and monitor the performance of model predictive controllers as well.

The only product I have found that handles both the regulatory control and the APC layer and supports a central server solution that I need is Matrikon. Do you know of any other product that can handle both ?

By the way, you are correct about the tuning package missig from ProcessDoc, but I understand it is now in beta.

Check the new stuff out at "www.matrikon.com/products":http://www.matrikon.com/products

Gino B

J

#### John Gerry

Gino,

Next month (February 2003) we will be releasing the next version of PlantTriage. It will have a connection to other historians - PI, InfoPlus 21 etc. via OPC HDA. HDA is Historical Data Access. It is a new OPC standard that soon all will embrace. You can read more about it here:

http://www.opcfoundation.org/04_tech/04_spec_hda.asp

So, with the new release of PlantTriage and OPC HDA you do not need to maintain another historian.

We are planning on DMC Monitoring and performance assessment for our summer release of PlantTriage. See http://www.expertune.com/ptNewStuff.html for the latest.

John Gerry
ExperTune Inc.

P

#### Pasi Airikka

Hello Jim!

I am quite late with my recommendation but better later than never...

One option for real-time monitoring and performance assessment of unit PID control loops
is LoopBrowser(TM) by Metso Automation Inc. (Finland). LoopBrowser is capable of providing control performance web reports (or in printed format) automatically or on user's demand.
It is, of course, a commercial product or to be more precise, control performance service. It is recommended to rather use it as a part of service provided by Metso Automation than just a stand-alone program.

For more information, visit Metso Automation homepage (www.metsoautomation.com) or the direct link to LoopBrowser and TuneUp (for PID controller tuning):

http://www.metsoautomation.com/auto...tion/EPprod.nsf/WebWID/WTB-010912-2256A-D28DE

Best Regards,
-P. Airikka

A

#### Andy Clegg

Dear All,

Following on from Pasi's comments - I thought it might be useful to highlight our web-site that has a small "Learning" section on control loop benchmarking. This can be found at:

http://www.isc-ltd.com/benchmark/

There is an on-line introductory presentation that I gave at a recent event, that might be useful:

http://www.isc-ltd.com/benchmark/learning_centre/aclegg.html

Hope this is of interest,

Andy [email protected]___

Advanced Control Technology Club, Industrial Systems and Control Ltd.,
50 George Street, Glasgow, G1 1QE Tel: (+44) 0141 553 1111
http://www.isc-ltd.com/actclub.html Fax: (+44) 0141 553 1232
______________________________________________________________________