Standpipe level indicators not reading accurate causing overflows 2

[itsleighton]

We built a water plant next to an existing standpipe that has an overflow level of 100.75'. We have installed a level indicator at the base of the standpipe converting the pressure to elevation and transmitting the level to a newly installed SCADA system. The high service pumps to the standpipe are programmed to turn off at 100' when the system is being run in auto.

Multiple times, we have witnessed these pumps fill up the standpipe to the programmed cutoff level when in auto and cut off exactly the way they are supposed to. For the last 3 nights, we have left the system in auto only to return in the morning and find the standpipe overflowing and the SCADA system indicating a level below its cutoff level of 100'. (1st time-99.5, 2nd-98.9, 3rd-99.0) To put it simply- the water is obviously higher than what the level indicators are saying it is. But it isnt doing this consistently.

All three times that we got an overflow, we have turned the system off, let the city use the water in the standpipe until the level drops enough to call for the high service pumps to turn on, we then turn the system back into auto and watch it work exactly the way it is intended. (no overflow condition). In other words, we cannot mimic the condition to witness it.

We lowered the cutoff level and all other lead, lag start / stops by 2' across the board and didnt have an overflow today, but I am curious to see if I will get a call in the morning or late tonight.

Information on standpipe: 35' diamater, with the influent pipe entering through the bottom of the standpipe and there is not a fill pipe running to the top (As the head pressure increases, the pumps serving it slow in GPM).

The level indicator is an Ametek

I have two questions.
1) Does the fact the influent pipe enters the bottom of the tank close to where we have the level indicator (on the side of the tank 3' from base) create the risk of turbulence and or "uplift" causing fluctuations in the pressure thus creating a wrong reading in the Level Indicator?

2) We have redundancy in the level indicator and this condition happens regardless of which one we choose to indicate system control. Has anyone had experience with level indicators operating with accuracy levels of less than 99.5%? My controls subcontractor says he has not- I personally have not previously.

Any ideas what is going on here? Ideas on troubleshooting? Solutions? Thanks in advance.

 

[QualityTime]

I think Item (1)of your questions is the key. I think you are getting velocity head influence on your readings. It is best to have instrumentaion located in a quiescent area. However you need to provide a sketch of to be clearer on what you think is happening

 

[danw2]

Nice post, lots of information.

What is the pressure transmitter's accuracy spec and what is its possible range (0-100psi?) and what is it ranged to (43.35 psi)? Is it a 2 wire, loop powered device?

>risk of turbulence and or "uplift" causing fluctuations in the pressure thus creating a wrong reading in the Level Indicator?

Yes, the indicated pressure/head can be affected by any positive or negative hydraulic force that impedes on the pressure sensor's diaphragm, but from what you've described, nothing changes between your visual observations when the system works and the overflow gremlin at night that causes a reading somewhat lower than actual.

It sounds the pump runs off a VFD, so presumably the flow rate is the same for any given level. One would think the turbulence pattern would match the flow rate, so why would the measurement change when the flow rate and its 'turbulence' is the same at a given level?

Can the location of the sensor physically change? Your "side of the tank' installation does not sound like a hang-from-top submersible that can move along the bottom and get closer to or further from the fill point if its anchor link has broken.

>Has anyone had experience with level indicators operating with accuracy levels of less than 99.5%?

Yes, but . . .

Head pressure is subject to errors due to density change, but ground water's density change isn't a factor here.

Submersible head pressure transmitters require an atmospheric vent tube for the reference side of the sensor. Over time, cool water can condense humidity in the vent tube (cool water is lower than the dewpoint of the moist air in the vent tube). The moisture can collect in the reference vent tube, drip to the bottom and apply a pressure to the sensor's 'low' side, causing a measurement that is lower than expected. But the vent tube is deadheaded, there's no 'flow' through it, so this tends to be a slowly occurring situation over a long period of time, and when it does happen, it doesn't reverse itself (moisture evaporating) from day to day.

Pressure sensors are subject to measurement uncertainties due to the temperature of the sensor. Industrial 2 wire loop powered devices are compensated over a very wide range of ambient temperatures, but is your Ametek? Is cool nighttime temperature making it read lower than southern exposure (in the northern hemisphere) to sunlight when it heats up?

>redundancy in the level indicator and this condition happens regardless of which one we choose

I assume that means that you've had overflow with either level transmitter?

The fact that overflow happens with either makes me suspect the other components in the system.

A level error can be from the transmitter or the other components in the system
- the power supply
- the analog input
- the SCADA signal conditioning calculations

A marginal DC power supply driving the transmitter's current output ( I assume these are 4-20mA current transmitters) can starve near or at the upper current limit, ie 20mA, which is where you're seeing your problem. This could be the situation if the transmitter's output signal is driving the SCADA's AI that has 250 ohms of resistance and a local indicator that has 250 ohms of resistance, and the transmitter is designed to drive a maximum of 500 ohms at 24Vdc. If the power supply can't hold 24V, and slips to 23.5Vdc, then the transmitter won't turn off, it just won't drive a full 20mA at its span value because there's insufficient voltage to do so. Is your 20mA value something close to, but above 100 feet?

Could the error be at the analog input? Do the 'redundant' pressure transmitters share the same input channel (switched by a relay)? Do the 'redundant' pressure transmitters different analog inputs on the same AI card?

If the the transmitters share a common AI input, then a faulty low order A/D bit could cause the error.

If they have different inputs on a common AI card then an intermittent common mode problem could shift the readings, or a shared amplifier/A/D on a multiplexed AI card could produce erroneous readings from both transmitters.

Multiplexed inputs frequently have low immunity to common mode voltage issues. Common mode appears as an offset shift in the readings. Some device other than the pressure transmitters that shares an AI card can cause common mode problems. Although common mode tends to be fairly constant, I once witnessed recorder signals shift when a 10000 HP motor was running but come back onto scale once the motor stopped.

The AI electronics are subject to error due to the temperature of the electronics. The error you see is larger than what one might expect due to temperature change of the AI but is someone closing the panel door at night but opening the panel door during the day? Does the air conditioner that is supposed to cool the electronics panel get turned off at night because 'no one is there'?

 

[itsleighton]

Here are some rudimentary drawings showing the best I can illustrate the field conditions.

 

[itsleighton]

Number 2

 

[itsleighton]

number 3

 

[itsleighton]

Thanks for your informative post. I am going to answer what I can and I have an email into my controls subcontractor. Im at home and dont have the submittals on this computer.

It sounds the pump runs off a VFD, so presumably the flow rate is the same for any given level. One would think the turbulence pattern would match the flow rate, so why would the measurement change when the flow rate and its 'turbulence' is the same at a given level?

These pumps actually do not run off of of a VFD. The MCC is just on / off. The pumps curves are such that when see see a higher level in the tank, the pumps begin to decrease their flow rates due to head pressure.

Can the location of the sensor physically change? Your "side of the tank' installation does not sound like a hang-from-top submersible that can move along the bottom and get closer to or further from the fill point if its anchor link has broken.

Yes, it possibly can. Let me know if you can understand what above pictures show in regards to the installation. It would require welding a new pipe the side of the tank and attaching a ball valve to the end of it, then drilling the standpipe "wet", then closing the ball valve once the penetration is done. I have a feeling this is what will have to be done.

I assume that means that you've had overflow with either level transmitter?

This assumption is correct.

A marginal DC power supply driving the transmitter's current output ( I assume these are 4-20mA current transmitters) can starve near or at the upper current limit, ie 20mA, which is where you're seeing your problem. This could be the situation if the transmitter's output signal is driving the SCADA's AI that has 250 ohms of resistance and a local indicator that has 250 ohms of resistance, and the transmitter is designed to drive a maximum of 500 ohms at 24Vdc. If the power supply can't hold 24V, and slips to 23.5Vdc, then the transmitter won't turn off, it just won't drive a full 20mA at its span value because there's insufficient voltage to do so. Is your 20mA value something close to, but above 100 feet?

Could this simply be solved by increasing the range the 4-20 drives? Instead of 102' being 20mA, make is 150'?

Could the error be at the analog input? Do the 'redundant' pressure transmitters share the same input channel (switched by a relay)? Do the 'redundant' pressure transmitters different analog inputs on the same AI card?

I believe they do. We are getting into the area where I need my controls guy to answer these questions. I am just an idiot pipe installer. I will update this post once I get a hold of him today. Thank you very much for your input thus far.

 

[itsleighton]

Picture #4

 

[danw2]

SCADA might have some logged level and motor event data to indicate whether turbulence/velocity induced error is the culprit.

>All three times that we got an overflow, we have turned the system off, let the city use the water in the standpipe until the level drops enough to call for the high service pumps to turn on, we then turn the system back into auto

Presumably "we have turned the system off" means you override the auto pump control and manually turn the pump off (while the level transmitter and SCADA remain powered up).

When the pump turns off,
- if the level stays at its faulty low level, and only drops gradually at whatever the consumption rate is, then the problem is NOT turbulence/velocity induced error.
- if the level moves fairly rapidly from its faulty (low) level up to the overflow level, then the rest of the elvel system is working and the influent turbulence is causing the erroneous low level.

Beware that the level response can be skewed by damping or averaging on the level signal either at the transmitter or in the SCADA.

My associate (who is from Prince Edward Island) thinks it's tidal error.

 

[itsleighton]

When the pump turns off,
- if the level stays at its faulty low level, and only drops gradually at whatever the consumption rate is, then the problem is NOT turbulence/velocity induced error.
- if the level moves fairly rapidly from its faulty (low) level up to the overflow level, then the rest of the elvel system is working and the influent turbulence is causing the erroneous low level.

Extremely good points. I hadn't thought of this while there. I will let you know what I find tomorrow. Thanks.

 

[QualityTime]

"...if the transmitter's output signal is driving the SCADA's AI that has 250 ohms of resistance and a local indicator that has 250 ohms of resistance, and the transmitter is designed to drive a maximum of 500 ohms at 24Vdc..."

1. Check to see what the SCADA's AI resistance actually is from catalogue sheets
3. Check to see what the local indicator resistance is from catalogue sheets

Compare the sum of the two abocve against what resistance the transmitter can drive through at 20 mA. I recently had a project where I had to boost the 4-20 mA signal becasue the signal from the PLC card could not overcome the resisitance in a valve operator. The valve operator would not settle in one position. It would reach the postion and just slightly continue to drive open then it would slightly reverse in the other direction and continue to do this. The only way I knew this was happening was to measure the 4-20 mA signal in and out of the valve and I could also watch the valve shaft moving

 

[semo]

I'm assuming that the transmitter is a transducer and not a switch.

Either way, you need to make sure that if the transducer is on the top of the pipe, there is a side blowoff to flush the line to the transducer. Probably need to with a switch also; but, I haven't seen problems with them.

A problem we've seen on numerous installations is:
Air can pocket in the vertical line beneath the transducer and give bad readings (air is compressible while water is not). Once the blowoff is opened and the air is purged, the transducer will sense correctly again.

Being on the side of the tank is the best setup and it is very doubtful that velocity/friction losses will affect your sensor.