Is it the end of PID & Model based controllers?

[DP-JYC]

I believe that: YES.
It's time for a new era: #RelaxC

I did port the brilliant algorithme of RelaxC from #AppEdge to EcoStruxure Automation Expert.

You can check the video over there: Youtube Link  explaining how it's working.

What to remember?

• No overshoots
• ~5 times faster
• Tunning with physical values
• With ONE block you can control every kind of processes

If you have any question, do not hesitate, I will be happy to answer.

🚀 Keep the control of your processes and stay tuned 🚀

[john_m · Accepted solution]

Hi,

I agree,  Relaxc  overcomes all the physical complexity of your process and more yet.

https://www.researchgate.net/publication/347510415_Relaxc_vs_Kalman 

[Bram-Marsman]

This is interesting, thanks!

I do feel the comparison in the video isn't completely honest, you are first stipulating a completely randomly tuned PID controller saying "look how bad it is", then you take Relaxc and do step-response analysis and tuning parameters calculation and then fine-tuning it yourself - you should show the same steps with PI(D) controller and run through Ziegler-Nichols or other methods...

Is the Relaxc algorithm something open, or is it proprietary from AppEdge and does every Relaxc controller in the future supposed to create royalties for your company? 🙂

[Grant_LeSueur]

Jean -Yves.

This does indeed look very promising especially from an ease of use perspective -

This could be very useful for those are less experienced loop tuning techniques.

I think your video  and maybe you CAT type could have benefited by adding an Integrated absolute error measurement.  This accumulator is a good way to measure control loop performance.

Additionally the presentation could have benefited talking a bit more about the algorithm.   One aspect customers really want to understand is unintended failure modes.  Are there scenarios where this algorithm could loose its mind?  It is for this reason customers don't really like to add and leave auto tuners on their process as they cant tolerate unexpected outcomes

You demo looked liked the natural process reaction time got faster when the performance comparison was made or you had too little Derivative action  in the PID example.  Derivative action used to be call ""Pre-act" because it acts first and should have driven the output to full-scale to get the process moving as fast possible.

I do like whoever that in EAE we can offer customers a choice to go with tradition PID approaches and or try alternatives.

IAE measurement is a measure of "Out of spec" which could be wasting raw material or energy. over time so its a good thing to improve.  Stability also means - less wear and tear on equipment.

[DP-JYC]

Sorry, I didn't saw your message @Bram-Marsman 

You are totally right, the example in the video is not fair.
I just took the "Demo application" of EAE (from the help), and yes the PID is not well tuned. Like a lot of loops in the industry are poorly tuned 😉

The issue is that a lot of people don't know much about Ziegler-Nichols, Naslin, stability criteria, robustness ... and when you know all the theory, you have to implement it and it's also not that easy with all IDEs / techno’s.

Yes also, #RelaxC algorithm is proprietary and we are building a business out of it 🚀 (I think that it's quite common to do business with your own IP 👍).

While we want to protect it, we run directly in the PLC in a matter of minutes. Not like Model based controller or cloud & AI solutions where you need months of data, going back and forth with engineers, Data Scientists going in various algorithms and getting some values… We want to give back the control to operators & people in the shopfloor.

✨ Stay tuned ✨

[DP-JYC]

Hi @Grant_LeSueur 

You got it right, ease of use is the key point 👨‍🏭.

Since the video we improved a lot of things to make it even easier, testing all modes and scenarios… Manu / Auto / Cascade / Multi Output like Hot Cold / 3 Layers / x layers / Limits / Output rate / Input Rate ( 🔥 Yes that’s a really hot & cool new feature that I have to speak about somewhere 🔥) / Tracking / … Because you are right @Grant_LeSueur , a loop is not only [PV / SP / algorithm / OUT], but it’s everything around and you have to cover all edge cases, failure, & co.

I take good note of the IEA measurement as it’s a good indicator, thank you for the advice.

I should put it for a next video! You want to suggest a complicated process?

You want to know more? do not hesitate to contact me, we will be happy to help.

BTW, since last time, we have now a website as well: https://relaxc.dynamicprocess.io if you want to discover more.

✨Stay tuned ✨

[Grant_LeSueur]

I like you new web site.

And you hit the nail on the head in a digram that visually shows IAE.  I also like your simple pricing model.

Do check your Hyperlink to your website below it launched my browser with bad characters like so - https://relaxc.dynamicprocess.ioÂ%C2%A0

https://relaxc.dynamicprocess.io

https://relaxc.dynamicprocess.io 

Grant

[DP-JYC]

📎 link fixed 📎

Thank you @Grant_LeSueur for all these feedback, really appreciate 🙏.