Help writing plc code for a bag filling process

That's my point why even consider a PID even if you strangle it to a P & limit it's range. However, as a learning curve a P controller is not a bad idea, whether it will work efficiently or not will depend on a lot of factors. but also on this point a P only controller will almost be like doing a simple bit of maths so as the error reduces so does the speed, again this leads to a long fill time.
I have been in this game for over 37 years now & would never use a PID in a batching process unless it was just to control a given feed rate not amount. often used in making Mayonnaise where oil addition has to be controlled at a given rate, the amount is always done either with load cells or flow meters so the PID is purely to get the flow rate right.

parky said:

I have been in this game for over 37 years now & would never use a PID in a batching process unless it was just to control a given feed rate not amount. often used in making Mayonnaise where oil addition has to be controlled at a given rate, the amount is always done either with load cells or flow meters so the PID is purely to get the flow rate right.

Click to expand...

37 YEARS?!!? WOW. I did not expect you guys to be this experienced. I mean i am just 22 years old. You have more experience than my entire life . So the standart on these things is just to give it different stages huh. Welp then i guess it shouldnt be any different for me but in the future i am definitely thinking of something like an ai. I dont know if its possible or not but i am thinking about it nonetheless

Yes you could engineer some sort of "AI" if you like, but is it worth it, will it produce any better results or even work, I have seen where in a batching process the in-flight was self adjusting, often the addition valves are above the vessel, this means there is a quantity of product in-flight between the vessel weighing & the shut off valve, it is common to use two valves, one is full flow & the other a trickle valve so that as it approaches set point the main valve closes leaving the trickle valve to dose the last little bit, this reduces the in-flight amount and often the previous batch over/under dosage is used to adjust the in-flight so as to reduce the error, however, often this does not work & can actually make it worse, the nature of a production plant means that water and some other products are fed to more than one process, the pressure therefore the flow can vary slightly and a sudden change in pressure/flow will knock out or make worse the in-flight, even if the feed is controlled by a PID a sudden opening of a supply to a tank will require a finite time to allow the delivery system to recover. the alternative to load cells on liquids is to use flow meters this reduces the real need for In-flight but as the vessel requires load cells for other products like veg or meat it becomes costly, you will find there is a trade off between accuracy, cost of plant, production speed & many other factors.
Very often Software engineers have to overcome the mechanical deficiencies with code as when commissioning production are on every suppliers back mechanical changes take time & invariably one supplier is late & the commissioning team are the last in line.

parky said:

Yes you could engineer some sort of "AI" if you like, but is it worth it, ...

Click to expand...

It depends on the application. If the product must be guaranteed to have a minimum weight of product, say 10kg, and the standard deviation of the system is 0.1kg (1-sigma), then, assuming a normal distribution, setting up the controls to target a 10.3kg mean (3-sigma over guarantee) would ensure 99.7% of the packages will meet guarantee, leaving three per thousand that could be rejected or could be returned.

Reducing that 1-sigma to 0.01kg means the target could be 10.03kg for the same reject rate, which saves 0.27kg per package on average, which will have some value. The magnitude of that incremental value is what determines the worth: if we are filling sandbags for volunteers building emergency flood barriers, the incremental value is probably zero; if we are filling bottles with a drink product, that incremental value is non-zero, but may still not offset the extra effort to design real-time AI optimization vs. a manual optimization every month or two.

Exxon used to sell the worst gas and make the most money because they had the best controls and could control closer to the minimum octane spec; other petro companies had to control to a higher average octane to meet the spec and gave product away.