Process Simulation - Digital Twin?

Hello Everyone, I hope we are all well.

I have a Standard System, (Rockwell - Compact Logix) and we have been building these for years. The system is a RO/EDI water purification skid.

It is heavy on Flow and Pressure PIDE controls.

I am being asked to make some drastic and unproven changes to how the system operates.

Does anyone have any experience with Process Simulation Software or Digital Twins that could help me prove out the PIDE controls before we build this monster?

Kind regards, P

I would advice using a HMI and PLC simulator.


I have built several "virtual" machines using Twincat/CodeSys to create a mock-up of all the sensors, pumps, switches etc. in the HMI.


I can then switch from the HMI "in/output" to the real in/output once the hardware is in place.


This is the "easiest" way IMO.


You could also go full on and simulate each sensor/pumps using math/curves in something like matlab/simulink https://www.mathworks.com/help/fusion/sensor-models.html


It really depends on how close you want your model to represent real life. I've been ok with using simple HMI input with sliders etc to observe how the system reacts.


To put you even more down a rabbit hole, look at TRL (Technology Readiness Level) https://en.wikipedia.org/wiki/Technology_readiness_level. You can use this as a tool to maybe get your team/manager to agree on what level you should be on before shipping to customers. Risk is high at the bottom but cost is low, towards the top Risk is low but cost is high.


Hope this helps somewhat

I also use a PLC programmed to simulate the system. I learned a lot more about the process in having to learn the science and maths needed to program the simulator. So, it depends on how complex your system is and how good is your maths. Using SI units keeps the maths simpler, though I still cheated in places by using averaging instead of maths to simulate how quickly pressures changed throughout the pipes.

BryanG said:

I also use a PLC programmed to simulate the system. I learned a lot more about the process in having to learn the science and maths needed to program the simulator. So, it depends on how complex your system is and how good is your maths. Using SI units keeps the maths simpler, though I still cheated in places by using averaging instead of maths to simulate how quickly pressures changed throughout the pipes.

Click to expand...

This is pretty key for a lot of things... and pretty interesting to do when you have data to create your models.

Just a few hours ago me and the chemist here on site were looking at his trend of the reaction profile versus flow and accumulated flow to work out the formula that creates this data. It's not really for simulation though, it's to be ran alongside the plant to compare a batch against the batch running to spot deviations sooner. (A bit like the golden batch implemented in historians, but with actual alarms and low on operator monitoring).

Thanks everyone.

Gets a little trciky when you have several CVs that can impact the same PV.

Another idea I am kicking around is building a scaled down system.