controller settings

Oct 04, 2006 5 Replies

Hi there, We are replacing around 300 old analog single loop controllers (F&P



4000 series) with digital equivalents. The implementation of the control algoritm in the new controllers is the same, so the gain/reset/rate are transferrable from the old controllers. However, the settings on the existing controllers are set by potentiometers, which we don't necessarily trust since the electronics have likely drifted over the last 20-30 years. So rather than take the old settings from the pots, we want to calculate them from the open loop response of the controller. I was wondering if there are any existing tools for doing this kind of test? I've searched around the web with no luck. Most of the software tools that would seem to do this look like they work through through the communication port of a modern controller (which obviously doesn't exist on the old analog controllers), rather than the actual controller I/O. Right now we plan on manually injecting & bumping an input, recording the output and then busting out the pen & paper to calculate the gain etc. Is there an easier way?! Thanks.

Quite likely, many of the old controllers were not well tuned, anyway. This is as good an excuse as you are ever going to get to tune these loops.

I think the method I describe in this tutorial is easier.

formatting link

Try this link:

formatting link
There is a pretty good little software tuning program once you have your transfer function from your open loop test. I don't know of any way it automating that without paying big bucks for something like experttune that would do everything for you...

Curtis

GbV

I think this discussion could benefit from some economic considerations.

300 new controllers, purchased and installed, with perhaps associated maintenance/upgrades to any failing sensors and valves, must be a project running $150K+ (in the process industries in the USA, it would be more accurate to add a zero to that number).

And then you are going to use hundreds of man-hours, consume feedstock, and make off-spec product, al so you can "save money on expensive tuning software?"

Spending $8-$10K on tuning software is a tiny fraction of the project costs and it will easily pay for itself before you have tuned two dozen loops.

That means it will be making you money while you tune the last 275 loops.

Check out

formatting link
to educate yourself on a range of pertinent topics.

Contact

formatting link
to learn about some proven, easy-to-use and reasonably priced software for data collection, process analysis and loop tuning.

Good luck with your project.

Doug Cooper snipped-for-privacy@controlguru.com

As you go through this conversion, be sure to consider the complete control loop. This includes ensuring that the sensor, the valve, and the controller are all working properly. Tuning is certainly important...but all the tuning in the world won't help a sticky valve.

I have been doing this kind of work in plants for 20 years. On average, I have seen:

  • 35% of valves have major problems (Stiction, hysteresis, poor sizing)
  • 85% of controllers are not properly tuned.

I have used a hardware A/D converter to bring 4-20 mA signals in to the ExperTune software. You can gather data and diagnose the valve, sensor, and controller, and determine optimal tuning.

Using software tools to optimize loops pays for itself very quickly. As mentioned in another reply, you are spending a lot of money to upgrade. Spend a small amount to save time and make sure everything works right...a very good investment.

Also, there are some tutorials and papers on loop optimization at:

formatting link

-George

I've a couple of products that should also help you:

formatting link
These are the tools I use in all my tuning work.

Regards John Greene Contek Systems Ltd Aberdeen

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required