packaging control with load cells and mitsubishi plc

hi i need your help
I have a Mitsubishi FX3U plc and screen GOT1000 16 ..
I have two Scales, and two load cells METTLER TOLEDO IND110

I want to automatically fill oil containers.
my load cells are well calibrated, give the correct weight. compare it with a digital scale

but the filling has much error. for example
the weight you should have is 18 Kg but always varies widely
18.2 kg, 18.5 kg, 17.5 kg, 17.4 kg. etc.

I want a more accurate measure and less mistake
and I think the problem is in the control algorithm in the plc

its not fast enough to close the valves in 18 Kg

I can accept an error between 18.05kg and 18.20 kg
a part of code is

[D>= D414 D901]-------------->(M229) valv1
[D>= D416 D427]-------------->(M230) valv2

D414 and D416 are the weight that is filling and D901 and D427 is the weight limit 18.10 kg

when D414 is greater than or equal to D901, M229 is activated and valve should be closed immediately

This is a tough problem, and many companies have worked many years figuring out routines to accurately fill bags and packages with a set weight.

The main problem is that if you wait until the scale sees the desired weight to start closing the valve, you are already TOO LATE and will always overfill, no matter how fast is the valve closing. The instant you start closing the valve, there is some falling material alreay between the valve and the package. This is material that has already left the fill hopper, but not yet been "weighed" because it is in free-fall!

Solutions to prevent overfilling:

1. Make the distance between the fill valve and the package as short as possible.

2. Use as fast-closing valve as possible.

3. Set up a "dribble" mode, a reduced material flow for the last 1 kilogram. Fill the first 17 as fast as possible, then close the valve to reduce the flow to a dribble to finish up the last 1 kilogram. If your vavle does not have a built in dribble mode, you will have to cycle it on-off rapidly to get a reduced flow.

4. As most packaging machinery now does, set up an anticipation or learning algorithim that also weighs the error for each package fill. Based on an average of the previous overfills, the unit anticipates the cutoff point to achieve a very close weight without overfilling. In other words, if the fill weight is 18 kilograms, and the previous 5 packages have averaged an overfill of 0.2 kilograms, then the machine should reduce the valve closing point by 0.2 kilograms and stop the fill when the weight is 17.8 kilograms.

i already have the three first solutions you are giving me, so i will try to do the 4th solution but i need a little help

i have another problem what if the package doesnt fill enough until reach 18 kg established weight? for example i have 17.78 kg
maybe you have an example explaining more detailed regarding the algorithim

I have another problem what if the package doesnt fill enough until reach 18 kg established weight? for example I have 17.78 kg.

Click to expand...

Possibly it has an "anticipation routine" that is closing the fill valve too soon. If you rincrease the amount of "anticipation weight" (the amount of weight WHEN the valve starts closing, then you should be able to get closer to 18 kg.

I do not have any examples of a Mitsubishi PLC routine to anticipate the future weight. I could work on writing one. I have only the Missubishi GX Developer-FX, version 8.25 PLC program. I do not have a GOT1000 screen. But before I do, I need to know more about your system, specially your scales.

I have two Scales, and two load cells METTLER TOLEDO IND110

Click to expand...

The anticipation algorithim should be inside the scales controllers, so that it is includued with your pachaging machine. That is the best place for any method to control the package weight. Any PLC routine will be too late to provide good control. If your scales do not include such a routine to close the valve before the final weight is reached, then if your PLC is controlling the scales, you might be able to addd such a routine - a rigged method that will not work as well as having the scales control the valve.

Maybe you have an example explaining more details regarding the algorithim.

Click to expand...

The method includes setting up some history memory locations, where you store the last 5 final weights of the package. Then, after each package fill, the scale controller (or the PLC) compares the average of these historical weights with the target weight.

Beginning Cut-Off Weight = Target Weight = 18 kilograms
Historical Final Weight Average = (W1 + W2 + W3 + W4 + W5)/5 = 17.78 kg
Deviation = Historical Average - Target Weight = -0.22 kilograms
New Cut-Off Weight = Old CutOff Weight - Deviation = 18 - (-0.22) = 18.22 kilograms.

Then it makes an adjustment to the actual "Cut-off Weight", the weight reading that is used to start the valve closing. The CutOff Weight adjustment can be made after EACH weighing, using a rolling average of the previous 5 weights. The goal is to gradually narrow the Deviation to 0, so that each bag weighs 18 kg. Normally the Cut-off weight will always be slightly less than the Target Weight, due to the free-falling material that has already left the valve (but not yet in the package).

If the final package weights fluctuate all over (from too heavy to too light), then your system is not good. Your scale platform may not be stable, could be vibrating or moving, your scale signal may be subject to electrical noise, material may be sticking and accumulating in your chute or on your scale platform, your load cells may not be accurate, and so on.

thanks your tips were very helpful ...
but I found that the main problem was the resolution of my analog module FX2N-4AD was the very slow
then i changed it for the FX3U-4AD which is much faster
at the end still had the problem of overfilling
I resolved by closing the valve before it reaches the established weight

Congratulations on getting it working!