Insufficient scan time

I think i may have a problem with one of our slc 5/05's. This is the deal.

We recieve a pulse output from a promass 83 flowmeter to determine flow and in turn, shut off of the valves and pump when the setpoint is reached. We were setup for one pulse per gallon up until yesterday when they decided to go to 1 pulse per pound for greater accuracy. I found that in prime conditions, using 8 lb per gallon as our variable density, we would get 1 pulse every 25ms. (300 gpm is our absolute max. 190-200 is normal)

By lowering the pulse width on the flowmeter from 100ms to 20ms i was able to achieve balls on accuracy. I looked at the plc scan time and it was fluctuaing between 10 and 20 ms.

My concern is that none of the pids that same slc control were active at the time. I am worried that once those pid's become active, the scan time will increase and my count will not be accurate. SO my options are a high speed counter card (which we don't have room for currently. 13 slot rack is full) or to slow down the pumps. (we have a high flow nut for a plant manager and he would run 500gpm if he could.)

You could look at using an STI.......

This might be just the application for a Discrete Input Interrupt (DII). You can configure one module in your SLC chassis to generate interrupts that run a single Program file when an input changes state. There are some masks, and filters, and other such things associated with it. This would assure you that you didn't miss any of the flowmeter inputs.

It might also slow down the rest of the control program, but if you are running PID equations it might not matter because they are solved relatively infrequently.

To offload this function from the SLC-5/05, I'd consider a MicroLogix 1000 (the little 10-point one) or MicroLogix 1100 (because it connects to Ethernet) and run your high-speed counting (Micros also have interrupt functions) for the flowmeters in there.

If you happen to have DeviceNet I/O, there's also the opportunity to run a DeviceLogix routine in an I/O block to do your flowmeter counting.

kurt.brinker said:

I think i may have a problem with one of our slc 5/05's. This is the deal.

We recieve a pulse output from a promass 83 flowmeter to determine flow and in turn, shut off of the valves and pump when the setpoint is reached. We were setup for one pulse per gallon up until yesterday when they decided to go to 1 pulse per pound for greater accuracy. I found that in prime conditions, using 8 lb per gallon as our variable density, we would get 1 pulse every 25ms. (300 gpm is our absolute max. 190-200 is normal)

By lowering the pulse width on the flowmeter from 100ms to 20ms i was able to achieve balls on accuracy. I looked at the plc scan time and it was fluctuaing between 10 and 20 ms.

My concern is that none of the pids that same slc control were active at the time. I am worried that once those pid's become active, the scan time will increase and my count will not be accurate. SO my options are a high speed counter card (which we don't have room for currently. 13 slot rack is full) or to slow down the pumps. (we have a high flow nut for a plant manager and he would run 500gpm if he could.)

Click to expand...

You may be able to achieve the accuracy with the meter, but the reaction time of the valves will determine your overall accuracy.

I come across this situation all the time, and try to convince plant engineers, that an accurate metering system is only as accurate as the amount of product that continues to flow while the valve is closing. If the valve closes too fast, a huge pneumatic hammer will occur .

I typically have the massflows setup to provide pulse width of no more than 10 pulses per second. In your case, based on the maximum flowrate how many LBs per second is the massflow measuring ?

Ian

at max flow 40lbs/sec, normal would be around 33lbs/sec

we normally overshoot 8-10 gallons (64-80lbs) but they can deal with that. If it was an exact overshoot everytime, i could compensate for it but sometimes its within 1gal and other times its up to, but not more than 10gal. We are filling tanker trucks so they don't have a problem with 80lbs more in there. The overflow depends on filter flow and what pump is being used.

kurt.brinker said:

at max flow 40lbs/sec, normal would be around 33lbs/sec

we normally overshoot 8-10 gallons (64-80lbs) but they can deal with that. If it was an exact overshoot everytime, i could compensate for it but sometimes its within 1gal and other times its up to, but not more than 10gal. We are filling tanker trucks so they don't have a problem with 80lbs more in there. The overflow depends on filter flow and what pump is being used.

Click to expand...

How about setting up the massflow for 1 pulse per 2lbs, or 1 pulse / Kg. That would be a complete pulse every 50ms,so setting a pulse width to 15ms should give the CPU time to capture rising and falling edges.

Also, the valve reaction time will determine the accuracy

Ian

i'll have to look into that. we use 2" and 3" solenoid valves with 1/2" nitrogen lines (unsure on pressure) for the valves so there isn't much lag when they close.

I use a dynamic preact for applications like this and have good success.

I use the following:

calculated quantity = delivered quantity + (flow rate * k)

where k is a constant, you have to find the correct value for k through trial and error, the larger the k is the sooner the valve closes. I usually start off with 1.0 and then adjust a little till I get it with in tolerance.

Then when calculated quantity >= desired quantity, stop material transfer.

kurt.brinker said:

i'll have to look into that. we use 2" and 3" solenoid valves with 1/2" nitrogen lines (unsure on pressure) for the valves so there isn't much lag when they close.

Click to expand...

Is there a hydraulic hammer?, or is the liquid transferred by pump, then the pump stopped when close to setpoint toi reduce the pressure?

Ian

curlyandshemp said:

Is there a hydraulic hammer?, or is the liquid transferred by pump, then the pump stopped when close to setpoint toi reduce the pressure?

Ian

Click to expand...

it all happens at roughly the same time. The pump, valve feeding the pump and the valve going to the fill rack are all disabled at the same time. Pressure drops extremely fast when the pump cuts off and we haven;t had any problems with it so far. The line after the pump runs around 80-100 psi if i am not mistaken. Think its causing premature valve wear?

Ken Moore said:

I use a dynamic preact for applications like this and have good success.

I use the following:

calculated quantity = delivered quantity + (flow rate * k)

where k is a constant, you have to find the correct value for k through trial and error, the larger the k is the sooner the valve closes. I usually start off with 1.0 and then adjust a little till I get it with in tolerance.

Then when calculated quantity >= desired quantity, stop material transfer.

Click to expand...

never thought about that. The only problem is that they like to see gpm on the meter and on the hmi. I could easily change the meter to read in lb's (right now it displays gpm and pulses in lb's) and then use a operator set variablem (density) to show a gpm flowate. The density also changes on a daily basis here. anywhere from 7.0-8.0 lb per gallon. I am assuming the calculation would be off depending on density. yes? no?

kurt.brinker said:

it all happens at roughly the same time. The pump, valve feeding the pump and the valve going to the fill rack are all disabled at the same time. Pressure drops extremely fast when the pump cuts off and we haven;t had any problems with it so far. The line after the pump runs around 80-100 psi if i am not mistaken. Think its causing premature valve wear?

Click to expand...

With that kind of pressure, you could have a two stage preact (cutoff) first shuts off the pump, pressure in the line plus the kinetic energy of the liquid will allow you to reach the final setpoint. Reducing the pressure should help the life of the pump seals.

Ian

If you are truly measuring mass (lbs. or kgs) then the density of the material doesn't matter. However, if you are using a volumetric flow meter and calculating mass, then density will affect your calculation. But shouldn't change the preact, as long as your flow rate is in lbs/min. When you change your calculation, the flow rate will follow accordingly.

its calculating true mass. this is alot of info to take in. thanks for the highly informative posts. i will read up and see what i can come up with.

Kurt, using an event input is the best way, but you can also use the IIM instruction to update the IO image table for your pulse input and then repeat the instructions. You can also use the same counter in multiple places in your scan. This way you can make sure you actually look at the physical input state several times in a scan (provided you do an IIM just before looking at it). Be aware that IIM adds a chunk to system overhead, but it can be an effective way to deal with certain things.

The following is completely acceptable and will function just as though the PLC were scanning faster for input I:1/0

           I:1/0

RUNG 25: ---] [--------------+-CTU-------------+-(EN)-

                             |COUNTER     C5:0 |

                             |PRESET     32767 +-(DN)-

                             |ACCUM          0 |

                             +-----------------+

.

.

.

 

Rung 125: ----------------------+-IIM-------------+-

                                | SLOT      I:1.0 |

                                | MASK      0001H |

                                | LENGHT        1 |

                                +-----------------+

           I:1/0

RUNG 126: ---] [--------------+-CTU-------------+-(EN)-

                              |COUNTER     C5:0 |

                              |PRESET     32767 +-(DN)-

                              |ACCUM          0 |

                              +-----------------+

.

.

.

.

Rung 225: ----------------------+-IIM-------------+-

                                | SLOT      I:1.0 |

                                | MASK      0001H |

                                | LENGHT        1 |

                                +-----------------+

           I:1/0

RUNG 226: ---] [--------------+-CTU-------------+-(EN)-

                              |COUNTER     C5:0 |

                              |PRESET     32767 +-(DN)-

                              |ACCUM          0 |

                              +-----------------+

Note that we are using the same counter address in all three cases. This detects off/on/off transitions in the pulse that may occur while the PLC is in mid scan.