Hitting consistent target weights

I have started what is turning out to be a difficult project. We are filling bags with a material and trying to hit a target weight, 60 lbs. What is making it difficult is the flow rate is not consistent, 15 lbs/s - 30 lbs/s, because we are using air to move the material. I am trying to calculate the pre-act or weight to close the pinch valve to account for the reaction time of the system and varying flow rate. I started using an ABG filter to filter the weight as well as calculate and filter the flow rate. I am then using the flow rate and change in flow rate from the ABG filter to calculate how much the weight will increase during the constant reaction time of the system. I am then closing the pinch valve the resulting number of lbs before the target.
I have made many adjustments to the reaction time parameter and the parameters in the ABG filter but still unable to hit a consistent weight.
I am open to any ideas on other filters or algorithms to hit the target weight consistently.
Any help would be greatly appreciated.
Thanks
Brent

I dont think you can fix this by filtering the error.

By the increase in weight you are getting flowrate.
You have to establish how much material you get after you close the valve.
If you cannot a linear or other simple equation for material vs. flowrate, you could try to generate a lookup table based on experimental data.

Aaother possibility, maybe you can dose at a slower rate when you are nearing the end of the dosing.
In that case even if there are variations, the differences will be relatively less.

You need to make the process less variable. The A-B-G filter does not smooth out the process/reality, just your estimated of what it is doing.

My suggestion is to have an infinitely fast measuring and shut off device so you can shut off at 60lbs. Since these devices don't exist I would try to get devices that are as fast as possible. If that doesn't work I would try to smooth out the flow.

The variability will be the flow as a function of time, times the time is takes to detect a trigger weight plus the time it takes to shut off the flow. Sorry, sometimes mathemagic will not help.

What is making it difficult is the flow rate is not consistent, 15 lbs/s - 30 lbs/s, because we are using air to move the material.

Click to expand...

You really need a Receiver Bin under the air flow, so the erratic and inconsistent flow can be converted to a very even predictable gravity flow out of the new hopper. Then use your methods to stop the flow slightly before the target weight is reached, so that there is a miniumum overfill. That only works if the flow rate is consistent (which it is not with air-powered product movement). Also, even with a Receiver Hopper, you should not attempt to fill and weigh bags unless there is some minimum amount of material collected in the hopper. Otherwise, you are still looking at the eratic flow from the incoming material.

Add a tall skinny hopper with a V-bottom, where you collect the material, then feed it out by gravity through your pinch valve. You may be able use the same air pressure from the air-conveying system to pressurize the top of the new hopper to make the material feed faster and with less clogging.

I have made many adjustments to the reaction time parameter and the parameters in the ABG filter but still unable to hit a consistent weight.

Click to expand...

I don't think you will ever get good results with your system as it is. As Peter said, there are some situations that cannot be corrected with math formulas and calculations. You have to change the filling system to make the flow rate smooth and precictable.

Air transfers are in consistent by nature. I did a hopper filler with air a few times and it sucks. You get slugs of material. We did what everyone else has said added a surge hopper and dosed to 80% then slowed down the dispensing speed. (LIMIT THE SLIDE GATE TRAVEL).

Air is not a good one, however make a big chute to prefill the bags with for example 50 lbs, or close to it, then move the bag to another fill station and top it with a smaller amount,
lets say you fill 2 seconds big you need 25 lbs/sec and 5 lbs/sec on the small one.
if you fill 5 seconds you need 10 lbs and 2 lbs.
etc.
filling times of 10 seconds are fast ones
if needed more speed another way of weighing is better.

only one scale is needed.
the proces gets a lot quicker and more accurate.

..make a big chute to prefill the bags with for example 50 lbs, or close to it, then move the bag to another fill station and top it with a smaller amount.

Click to expand...

That sounds logical, but how many operators are going to like the idea of handling the bag 2 times and doing about twice the physical work to get 1 bag full?

Topping with a slower fill rate is a common method, but is usually done by just installing or creating a dribble mode with the bag in the same original fill position.

Agree with what others have said, I just commissioned some powder delivery systems last year and the key is regulating the powder feed into or out of the air delivery system.

In my situation we regulated the feed of powder into the air delivery system with rotary valves which were controlled by a VFD. This allowed me to create fast feed delivery and slow feed delivery speeds and control the flow rate better as I approached the set point. With this type of setup we were able to achieve a delivery variance of 0 - 1% of the intended set point. Customer spec was 5%.

Sounds like this is a situation where the Controls Engineer has to sell the fact that the system is only capable of what you are currently achieving and that the physical traits of the system need to change. Best of luck!

This sounds like the classical situation of "can't the program fix it"?. Currently the mechanics of your system are not capable of delivering a steady rate. With an unpredictable rate, you're dead. You will pull out your hair to program it to work - and at times you will have that AhHa! moment when you think you succeeded - only to see it fail again a short time later.

As others have eluded to, you need to change your mechanical system. A receiver hopper or rotary valves on VFD or something. Unless you have a means of catching the powder in an intermediate holder or controlling the metering rate into the air stream, you're in for a long and frustrating time.

Thanks for all the comments, they are exactly what I was afraid of. Just to clarify there is a holding hopper with a spout on the bottom that the bag goes over. The air is to fluidize the material and help blow it to the bag. We can turn the air off and it will gravity fill most of the time but it is slow and sometimes will just stop.
This was also a retrofit from a mechanical balance beam scale with pneumatic switches, and we did the conversion because of accuracy issues. At this point we have neither gains nor losses in accuracy. I will now look more closely at the mechanical end to what I can do to improve the flow rate.
Thanks again for all your help.

The air is to fluidize the material and help blow it to the bag.

Click to expand...

Do you have an accurate pressure regulator to keep the air pressure in the surge hopper constant as the bag is filled? Material loss in surge hopper = pressure drop = changing material flow rate = uncontrollable bag weight.

babones said:

...We can turn the air off and it will gravity fill most of the time but it is slow and sometimes will just stop.

Click to expand...

What about switching of the air and use a vibro feeder to dribble the last grains?
These are pneumatic variants:
http://luftvibrator.dk/produkter/pneumatisk-vibratorer

EDIT:
Sorry, not your language. Look here:
http://www.vibco.com/products/pneumatic-vibrators

Kalle