Tuning Part 1 – The myth: ‘Tuning can solve all problems’
Basic principles and methodology of tuning were covered in Part 1 of the Loop Signature Series (available on CD for persons outside southern Africa). This and later articles will deal with other aspects and thoughts on this subject. The reason I am now presenting these articles is because I frequently get questioned about tuning, which is a subject of fascination to many, if not all, people in the control and instrumentation world.
One of the most frequent remarks I get from people is that they wish to come on a tuning course, and seldom that they would like to learn how to optimise control loops. I always remember one particular senior C&I technician from one of South Africa’s largest processing companies, who arrived at the start of the course and then took me aside and asked me why he had to waste a whole week sitting on a course learning to tune a controller. He said that surely I could teach people how to do tuning in a morning! My standard reply to such remarks is that he (or she) is attending the wrong course. The course I give is on control loop optimisation, of which tuning forms one of the smallest components.
Back to basics
The problem is that due to the lack of knowledge of the principles of practical control, as opposed to the teachings of theoretical control taught in the vast majority of educational institutions, there is a huge fallacy existing in plants worldwide that any problem can be fixed by tuning the controller. This not only exists amongst C&I disciples, but throughout the plant, from operators, through process experts, and right through to management. As soon as a control loop stops working nicely or starts giving problems, the cry rings out: ‘Tune the controller’. And then someone from the C&I department is sent out to do this. In most cases they play around with the magic P, I and D settings by trial and error hoping to make things look a little better.
In many plants tuning is considered a low level discipline, and left to instrument mechanicians and even in some South African plants, people like mill-wrights, and in most cases no scientific method of tuning is used, and most important of all, no or very little effort is made to analyse the loop and to find out why the control is not working.
Very few people stop to consider why the control has stopped working or is no longer operating effectively. It should be fairly obvious that it is extremely unlikely to be the tuning. How could it have changed? Tuning in the modern world is just a set of numbers stored in registers in a computer, so, unless someone has been fiddling, or perhaps the computer has been damaged by a surge, then the values should not have changed. However, if the control was previously satisfactory and is now no longer working properly, then something else must have changed.
Now what?
If this is the case then one needs to investigate what has changed and why this has happened, before even thinking about playing with the tuning. This takes us to the subject of loop analysis.
Although it is extremely difficult to completely generalise on the steps that should be taken, which can vary from case to case, the following are the procedures which are most often followed by practitioners experienced in the practical aspects of loop optimisation:
1. Understand the control system:
* Gain understanding of what the process is and the purpose of the control. Also ascertain the control response required, as not all controls are required to operate quickly. In fact some must deliberately operate extremely slowly. If any doubt exists consult a process expert.
* Review the drawings and see that the control strategy is correct, and if it is the best and simplest way to satisfy the control requirements.
2. If this is the first time you have worked on the loop, go out into the field and check that the installation is correct.
3. Check transmitter calibration and valve stroking are correct.
4. Check if there are any filters (damping) on the transmitter or in the control equipment. (I am very often told there is no damping, only to find later that this was incorrect.) Damping will probably have to be removed later.
5. Ascertain from the operators and process people what problems they are experiencing with the control or controls.
6. If several loops are involved make a sketch of the loops to really understand the workings of the whole system and to judge if there are any control strategy problems and what loops may be interactive or have an effect on other loops.
7. Ensure that you are familiar with the control equipment and that the control block has been programmed correctly. (I find that in about 85% of all PLC control systems, control blocks are set up incorrectly.)
8. Plan the testing strategy, for example the sequence of testing of the various loops. For example, some loops need to be in automatic when other interactive loops are tested, so they should be sorted out first.
9. On each loop one should preferably firstly perform a closed loop test with the existing parameters. This can help determine:
* Class of process (integrating or self regulating).
* If noise on the PV signal could present a problem (ie, cause the valve to move around too much).
* If there is any cycling and, if there is, to find the reason for it.
* If the control response is meeting the requirements.
* If the tuning is robust (safe) enough if changes should occur.
* Possible problems like poor control strategy, valve hysteresis, non-linear installed valve characteristics.
* Process stability and effects of load changes.
10. After this remove any damping or filtering and perform an open loop step test with sufficient steps to enable one to determine the following:
* Confirm the process class.
* Check that installed characteristics are linear.
* On self-regulating processes determine process gain to see if valve sizing and transmitter spanning are within reasonable limits.
* See if positive or negative hysteresis is present, and if it is, then measure it, and try and determine the cause, such as stiction, backlash, undersized actuator or positioner problems.
* Control strategy and process design problems.
* Process dynamics to enable one to determine the correct type of tuning to apply to match the control requirements.
* If noise will be a problem.
* Equipment non-repeatability.
* Load change effects.
* If interaction exists and how to treat it.
* Need for advanced techniques like feed forward or decoupling.
* And finally how and where to tune. This comes right at the end.
11. The tuning should then be performed using sophisticated scientific tuning software that works properly, and after that a final closed loop test must be performed, with the new tuning parameters, to check that the desired control requirements are now being achieved with sufficiently robust response.
I would suggest that 99% of C&I practitioners if they are completely honest and who have not attended a course on the practical aspects of control, will admit that they have little idea on how to follow all the steps in the procedure as described above.
This is a long way from going out and playing with magic knobs to sort out the problems.
Michael Brown is a specialist in control loop optimisation, with many years of experience in process control instrumentation. His main activities are consulting, and teaching practical control loop analysis and optimisation. He gives training courses which can be held in clients’ plants, where students can have the added benefit of practising on live loops. His work takes him to plants all over South Africa, and also to other countries. He can be contacted at Michael Brown Control Engineering, +27(0)11 486 0567, [email protected], www.controlloop.co.za
| Email: | [email protected] |
| www: | www.controlloop.co.za |
| Articles: | More information and articles about Michael Brown Control Engineering |
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version