feasible? self adjusting dosing valves

In work I have a machines that produces sachets of a product. 2 machines, 10 lanes each, 1 sachet per second (that's a total of 72,000/hour for those wondering)

My issue is that one sachet from each lane has to be weighed every 30mins. This requires an operator from each machine to remove 10 sachets (upsetting the count) and manually weighing the sachets on desk scales then climbing over the out feed belts and box conveyors to adjust manually adjust the dosing valve (pneumatic diaphragm valve with screw seat to adjust product weight) to within the 12.3-12.6g target.

What I would like to do...

Stage 1:
Automated dosing valves with the use of a few buttons and a stepper motor.
Positions required:-open, closed, +/- adjustment.
Safety issue sorted

Stage 2:
Unfortunately the scales that are used are not networked.
Build scales that can record data on an excel spread sheet.
Set lane on scale to record weight to column on spreadsheet
Set 30min time frame in rows
Reduced paperwork

Stage 3:
Weigh scales built into coveyor
Has this been done with adruino?
Close automatic weighing could produce an LCD read out for this stage for an opertor to adjust as required

Stage 5:
Make it all talk.
You never know this could mean that apart from calibration these 2 machines could be run by one person like they should

So...
Am I dreaming??

I'm sure this is all available off the shelf, but for a lot of $$$. No Arduino necessary. The really doubtful part is the practicality of weighing something on a conveyor with any reasonable accuracy.

actually easier than it might seem. say you take one out of the line, weigh it, then put it back. but you take one out and replace it with the one from the last count. then you never loose count.

as for the weight bit. more data is needed. needle valve ? pressure ? viscosity ? the thing that is adjusted if the 'final control element' and that needs to be able to be very accurate. my target has always been 10 times more accurate than the process requires, and a minimum of 4 times more accurate.

if you have a window of acceptable values and you set it in the morning for center of window and it creeps during the day, you could have an automatic adjustment to bring it back to center. ideally never exceed the boundaries.

but this also means good measurement.

so the answer is YES, it can be done.
the result is based on some factors,
are you willing to modify the machine to add this feature ?
are you able to finance the features ?

another thought is that you can put a motor on the adjustment and the operator can make adjustments from a distance.

The quote function doesn't seem to work on my phone

Aarg - yes $$$ is the main issue here as its a fairly new factory money is a very large factor as the current system 'works'.
Check weighing is pretty common in the food industry but again accuracy comes at a cost

Dave-in-nj - I like your idea of one in one out but there isn't room without serious modification to the machine which isn't feesable until all is proven ok
The weight valve is currently doing its job and us accurate. But the control falls down to human control were a 90degrees can make 2g adjustment. I would like to use a stepper motor to control this so the operator knows 1 press of the '+' button will turn the valve 2 degrees which relates to x in weight

Financing comes down to my manager deciding its worth I guess. I'm willing to prototype the automation side out of work as I enjoy it but the mechanical side is out if reach at home and will be very machine specific.

Yes, the operator station needs to have the controls, not operator running and climbing to make adjustments. Doing that alone should make the current process much better.

If the packaging always weighs the same then weighing a whole box should provide an average weight, no?

GoForSmoke - in theory yes but the reality is the weight changes regularly due to:
The pressure changing in the supply line when one machine stops
Sachets dont always make it into the box for various reasons
There is 250 within 4mins 10 seconds in a box and a tolerance of 0.3g. 1 too many or few Sachets would make that box a out of spec if calculated by total weight which would require manual correction.

Also if weighed per sachet in line the it could be linked in to stop the machine when there is no weight detected in a (normally very messy and time consuming) jam up.

elltee:
GoForSmoke - in theory yes but the reality is the weight changes regularly due to:
The pressure changing in the supply line when one machine stops

Which you do not want to adjust the valve because of. BTW, I know a bit about QC from my time in production industries writing apps that I ran the machines for, sometimes just to know what to write!

Is your air pressure from a big tank with compressor? Perhaps separate air lines from the tank to each critical use would diminish or eliminate that source of error? Would the cost be worth it in a year or less? Perhaps a small air tank (liter volume) with in-valve on the machine itself would buffer small changes in the line?

Sachets dont always make it into the box for various reasons
There is 250 within 4mins 10 seconds in a box and a tolerance of 0.3g. 1 too many or few Sachets would make that box a out of spec if calculated by total weight which would require manual correction.

Also if weighed per sachet in line the it could be linked in to stop the machine when there is no weight detected in a (normally very messy and time consuming) jam up.

Most every way I can think to check the mass of a sachet is not error proof. But you might look into how some inkjet printers used piezos to pump finely measured ink to paper.

With making adjustments, if you get too closely coupled to a system that has random effects then you end up making things worse.

GoForSmoke:
If the packaging always weighs the same then weighing a whole box should provide an average weight, no?

But each lane is separately adjusted, that won't work unless each lane goes to separate boxes.

Consistency is paramount. if one machine goes off-line, the result should not be that anything else is effected.

if you are filling bags with water, you could add an 11th line, then measure that and discard or re-cycle.

there are different types of control. you could just use pre-set points.
10 units running, set to 7
9 running, set to 6
8 running, set to 5

If over level is not crutial. like a few extra potato chips in a bag, then this could work.
if under/over is harmful, such a a pre-mix that has exact proportions, then the concern is higher.

would it be possible to measure flow and pressure at each station ?
would it be possilbe to dose into a measure, then into the final container ?
that would offer instant results and data logging for every unit, not just random batches.
Ultimately, measure each item is the final result.

With plastic injection-molded there is a cavity in the injector that gets filled to a set amount then injected under 2000psi. It gets filled through a 1-way valve and lead screw (plastic is pellets that melt before injection) in the back, linear feed->fill.

The current valve does it's job very well and doesn't need replacing. Its the human error that is the issue here as fine adjustments are required. I think the adjusting screw coupled to a stepper motor will cover all that is needed at stage 1.
Also changing a valve that works in a process that 'works' is never going to get approval from those above.

Due to the way the sachet is formed and filled the weighing is best done once it is sealed and leaving the machine.

Air is sterile atmospheric pressure the product line pressure is know to change.

I've come to the conclusion the conveyer weighing just doesn't seem likely with 0.3g tolerance as the belting will probably have more weight change in a full rotation

elltee:
Due to the way the sachet is formed and filled the weighing is best done once it is sealed and leaving the machine.

If the sachet package (not counting filling) is always the same size and weight then shoot them with a puff of air off a ramp. The lighter ones will go faster and fall on a different trajectory from the heavier ones. Collect the ones in the middle, when sachets show up in the high or low hopper the controller can detect and auto-adjust and/or sound buzzer and blink lights.

Or maybe a very small puff could move a sachet from one edge of the belt towards the other and collect what comes off the belt in different hoppers covering different sections of the width.

If your sachets could roll, vending machine coin mech tricks could be used.