Readers occasionally e-mail Michael Brown with questions after reading his articles. Michael felt - and I agree - that by publishing some of the discussion, it would probably be of value to others who have similar control loop related problems that they are looking to solve. Here is one such exchange that revolves around knowing just how far open (or closed) a valve really is. - Ed.
Richard Mackay writes
Sir, reading your article about non-linearity of loops reminded me of an experience I had last year tuning the density control loop for a DMS plant at an old diamond mine near Pretoria.
We got the water addition valve stroking OK: Closed, 25%, 50%, 75%, 100% ...all checked out OK when looking at the I/P converter controlling the valve. But when we tried to control on density - which is a slow loop (as you can only add water, not remove it) - we found that PID control was impossible.
I guessed by looking at the trend that the flow was only starting when the positioner was far from closed. After much acrimony I tested the valve the next day and found that it only started flowing at 25%!
On calling the valve supplier, his response was that they can change the position of the seat of the valve. I do not know why a proportional valve would require this feature but I have learned now not to trust the positioner to inform me of the flow through a valve.
Michael Brown's response
I am not an expert on valves, and I am not sure what you mean when you said that you judged the valve position by the I/P converter output, and then you also talk about judging the valve position from the positioner.
Are you referring to looking at the little pressure gauges on the valve positioner combination? If so, then I never trust them. Generally one can get an idea of what is happening with valve position by actually looking at the valve stem. However this is also fairly rough, and as you did, you can judge best what is going on by the process response, which is easy to do if you have a measurement of the flow through the valve. This also allows you to check installed flow characteristics through the valve.
Judging what is happening in a valve on a process like a DMS density, would be relatively easy to do if the water addition is on the pump suction before the density measurement. However, many of the mines add the water into the tank which feeds the pump. This is a bad thing as it turns the process into a very slow integrating process, and it takes a long time to correct density variations. If this was happening on the mine where you were working then it would indeed have been very hard to judge what the valve was doing from the process response.
I have just had a similar problem with DMS density control, and have just come back from a diamond mine where they were adding water into the tank via a splitter box that was set up completely incorrectly. Control was almost impossible. The water addition flow was going from zero to full from about 48 to 57% of the controller output range, making the control almost ON/OFF.
For more information contact Michael Brown, Michael Brown Control Engineering, 011 486 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