Vibrator Speed Control using Load Cell weight

[drbitboy]

Would my PV in my PID be the rate of my constant feeder or the rate of the feeder im controlling with my PID?

The latter: the PV in the PID will be the measured flow rate of the feeder controlled by the PID.

 

[parky]

I thought about asking a few questions like was this bulk mixing in which case yes you could adjust the ratio based on added (total loss of both products accumulated), but decided not to bother so many questions really & as I stated before PID may not be required.

 

[Cody Anderson]

Yeah so its a really large bin of peanuts that are being blended together so the moment to moment rate is not as important. Thats why I wasnt sure if a PID would be necessary. So there are actually 4 total stations each has a bag of peanuts hanging over them. Each has their own load cell. only two stations are running at a time, when a bag empties it switches over to the next station and continues to run while the other station gets a bag change. So there is no need to stop to change out the bag.

 

[drbitboy]

@Cody: hold on a second, I am going to re-define the terms here.

1. Two peanut flows
   1. One flow [hi], of the higher-percentage peanuts, is controlled by operator HMI sending a typically constant [percentage of maximum speed] directly to the vibrator
      1. That speed percentage is typically near 100%
      2. This flow is only measured
         1. the measured flow is Fhi
         2. For reference, this was formerly called Fmeas
      3. This flow is not controlled by PID, so there are no SPhi, PVhi, or CVhi values.
   2. The other flow, [lo], of the lower-percentage is controlled by a PID to ensure a target ratio of this flow in the total flow
      1. The measured flow is Flo, which is the same as the PID Process Value input, PVlo
      2. The setpoint of the PID is SPlo
      3. The controlled value output of the PID is CVlo
2. One ratio, PCTlo
   1. The weight fraction of the lower-percentage peanuts, expressed as a percentage
   2. Typically 10-20%
3. The lo flow's PID setpoint formula
   1. SPlo = Fmeas * PCTlo / (100 - PCTlo)

 

[Cody Anderson]

Okay,

Think of it as Station A, B, C, and D.
A and B have small peanuts
C and D have large Peanuts.

All stations have a bag on them to start. I set which stations are to match at the HMI. And I can control what percent each station should run at.

Station A and C will start to run either 10 and 90 percent or 20 and 80 percent. Can even go up to 30 and 70 percent.

Station C will run out of peanut much sooner than station A. So C runs out, and the program switches to station D to continue running large peanuts while station c gets a bag change. Then D runs empty and switches back to C.

Same thing for Stations A and B.

 

[parky]

You only need to select the one percentage (I think) so perhaps you select the higher value because the total will always be 100%.

 

[Cody Anderson]

So could I just look at it this way. If I run the larger peanuts at max speed 60 hz and always keep that constant. I can calculate my flow after running for a minute by taking my weight loss. Once I get my flow of one vibrator running at 60 hz I can just add 10 percent or 20 percent to that number and that would be my flow I am trying to reach with my other vibrator correct?

So say I am running at 12,500 with one vibrator at 60 HZ. That is just large peanuts so if I want an actual 90/10 ratio of large to small. I would need to take 10 percent of 12,500 and add it to that 12,500 and that would become my actual flow I want to reach?

I am just not sure how to create a scale factor and adjust the speed as needed on my smaller vibrator to speed up or slow down as needed.

 

[drbitboy]

Okay,

Think of it as Station A, B, C, and D.
A and B have small peanuts
C and D have large Peanuts.

All stations have a bag on them to start. I set which stations are to match at the HMI. And I can control what percent each station should run at.

Station A and C will start to run either 10 and 90 percent or 20 and 80 percent. Can even go up to 30 and 70 percent.

Station C will run out of peanut much sooner than station A. So C runs out, and the program switches to station D to continue running large peanuts while station c gets a bag change. Then D runs empty and switches back to C.

Same thing for Stations A and B.

C and D are hi: their rates are measured (Fhi); their vibrators are controlled directly via HMI; they have no PID.

A and B are lo: their rates are measured (Flo); their vibrators are controlled via PID as output CVlo; Flo is the PV input to the PID; SPlo is the SP input to the PID, calculated as noted earlier i.e. SPlo = Fhi * PCTlo / (100 - PCTlo)


Whether the process uses two PIDs, one for A and one for B, or uses one PID and multiplexes the various inputs and outputs, is up to the programmer; it will be somewhat messy either way, although using an AOI/FB/FC might make it cleaner. With one PID comes the assumption that the A and B vibrators have similar performance; with two PIDs comes the need for a second set of tuning parameters.



Obviously it could go hog wild, detect empty bags, automate the bag changes or detect a missed change, startup, auto-manual transitions, etc.

 

[parky]

I would do it this way you have a 4-20ma signal to an inverter running the large nuts, assume you have checked the range i.e. o-14,000 lb/hr then you scale the analog output to that so at 20ma it runs the feed at 14,000 lb/h.
The operator types in a total feed rate of 12,500 so this becomes the set point for the large peanut feeds, the PLC takes this 12,500, converts it to the raw analog value lets assume 0-32767 so if 14,000 is 32767 then for every pound there is 2.34 counts for every pound so times it by 12,500 minus the 1,250 (10%) this gives a raw count of 26,330 raw value to send to the inverter analog card.
Then take the 10% (1,250) and scale it for the second VFD & do the same thing.

Things to do:
Get the loss in weight from the two bags (cyclic i.e. every 5 seconds say) to produce the rate per hour.
Get the operator line speed (lbs/hr)
Get the operator percent of main addition
Do the maths, the setpoint for large peanuts will be the actual % of the line speed and the small peanuts will be that minus the actual.
scale the large peanuts feed speed into the raw for the analog card of the large peanut VFD & if using a PID then the small peanut lb/hr is the SP of the PID & the Actual is the small loss in weight. so in effect
the large peanut feed is running at 60hz (11,500 lb/hr)
The small peanut feed is running via the PID at around say 15hz but modulated by the PID algorithm.
I will see if I can dig out an old one I did some years ago, it will not be RA but you will get the idea.

 

[drbitboy]

So say I am running at 12,500 with one vibrator at 60 HZ. That is just large peanuts so if I want an actual 90/10 ratio of large to small. I would need to take 10 percent of 12,500 and add it to that 12,500 and that would become my actual flow I want to reach?

It's actually one-ninth i.e. 10 / (100 - 10) = 10 / 90 = (12500/9) / 12500, because the 12500 large is supposed to represent 90% of the total flow.


If the smaller peanuts were 1250, then the large/small ratio would be 12500/1250 = 100/10, not the desired 90/10, and the total flow would be (1250 + 12500) = 13750, so the percentage small would be 1250/13750 = 9%+, not 10%.

I am just not sure how to create a scale factor and adjust the speed as needed on my smaller vibrator to speed up or slow down as needed.

Please re-read at my first post in this thread: the PID takes a setpoint (SPlo, in this case the smaller-percentage peanuts' flow rate necessary to achieve the target ratio, calculated as noted earlier), and that PID adjusts its output (CVlo, the speed percentage sent to the smaller vibrator) to make that PID's process value (PVlo, the measured flow from the bag above the smaller vibrator) equal to SPlo.

 

[Cody Anderson]

Thank you guys! I am working on getting this setup in studio 5000 right now.

 

[parky]

Here is some simple code to get the ratios so on the HMI there is only fill speed & the main product speed in lbs/hr

 

[parky]

Things that come to mind:
1. How do you determine when the bulk bag is full ? is it also on load cells or are you going to use the loss in weight of the supply bags to give you the amount in the bulk bag.
2. Think about the changeover, when a bag becomes empty, it will probably not be zero, so think about the changeover.
3. Although the large peanut bag will be set to run at a given speed maybe it would be better to use the loss in weight from that bag to control the small peanut PID, so instead of using the setting calculated at 90% use the actual from the loss in weight. for example in that simple logic I posted, the flow rate of 12,500 lbs/hr is computed to 11,250 & the result of 10% is calculated as 1,250 this is fine but assume although the setting for large peanuts is 11,250 it may not be running at that (things like oil build up on vibrator, slight loss of motion due to mechanical wear etc). it might actually run at 11,100 for example, so by using the loss in weight flow from the large peanut bag to calculate the small peanut flow rate will mean the ratio will be closer. so if the large bag flow varies the small bag varies in unison, this may not matter if your accepted limits are within the tolerance of the mechanical restraints.

 

[drbitboy]

Things that come to mind:...

Although the large peanut bag will be set to run at a given speed maybe it would be better to use the loss in weight from that bag to control the small peanut PID,

I am pretty sure @Cody is doing this already:

So could I just look at it this way. If I run the larger peanuts at max speed 60 hz and always keep that constant. I can calculate my flow after running for a minute by taking my weight loss.

To get 10% small peanuts in the bin, the only issue here is the details of the math, i.e. the small peanut rate should be calculated as one-ninth of the available measured large peanut rate, instead of being calculated as one-tenth of the total rate because that total rate is neither directly available nor measured.

 

[brinocerous]

The PIDE instruction in Studio 5000 has a ratio mode that would be good for this application. See page 13 of this Rockwell paper: https://literature.rockwellautomation.com/idc/groups/literature/documents/wp/logix-wp008_-en-p.pdf