Sump pit with redundant level sensors

[rta53]

We have installed a sump pit at our facility that will be collecting water from floor drains and other sources. The pit has 2 pumps that will operate based on tank level. The water is pumped into a neutralization tank where it is neutralized before being sent to municipal drain. I was asked to install redundant level transmitters. I was wondering how any of you have done the programming for this type of application. My first thought was just to parallel the logic for each of the level sensors and if one fails the other will just keep things trucking along. Another thought was to use a copy or move instruction and use the data from one sensor and basically have the other in standby. Then if the first one fails I would start moving/copying data from the second into my level logic. The sensors can be programmed for failure modes of hold last state, empty, full, overfill(21mA), or overfill(21mA). I thought I would use one of the overfill failure modes to indicated a lost signal. The sensors we are using are ultrasonic transmitters with 4-20mA. I am using RSLogix 500 and our plc is a SLC 5/05. I have done many level applications but typically there is only 1 transmitter. Doing the redundancy thing is probably fairly typical in industrial applications but it's a new one for me. Any suggestions would be appreciated.

 

[OkiePC]

So if one sensor says the pit is full and the other sensor says it's half full, what should happen?

Is it worse to have the pit overflow or have a pump cavitate?

Can you use the rate of change of the signal as verification of a good signal, so when one of them jumps suddenly you can consider it a bad input, and ignore it, keeping the pit level under control using the assumed good signal?

Also, be sure to make use of the open circuit bits in the SLC as added determination of a failed sensor or wiring.

It might also be wise to hardwire a selector switch for each pump contactor to bypass the PLC altogether so in the middle of the night, Bubba can prevent a flood. We have a waste treatment system with a pit and this manual bypass has saved us from having the waste overflow into the city sewer line more than once.

 

[rta53]

We do have hardwired hand switches that bypass plc control. That was a given in this application. You make a good point about what to do if the sensors have different readings. It would probably be better to have 2 different types of level indicators. Maybe some type of float switch would be an option for backup. Perhaps it would be better to just use 1 of the sensors and bring the other online if the first one fails.

 

[OkiePC]

If I were using two analog sensors, I would constantly calculate the difference and if it exceeded a tolerance, I would execute logic that made the decision which one to trust. If they were within tolerance, I would just use one of them or the average of the two for control.

If you used float switches as backups, you could check the scaling of your analog sensor(s) each time the float switches changed states.

Our pit uses a capacitive probe made by Drexelbrook with a special transmitter that feeds 4-20mA to a process display with relays. It seems to be pretty reliable, but the acids in the animal fat and the cleaning solutions it encounters do occasionally lead to failures. I should probably consider a radar, laser or ultrasonic here too.

 

[Gil47]

What has created the situation that they don't trust the one they already have.

As the sensor is ultra sonic , from personal experience they don't like foam on the top of the liquid, they don't like the pit to have alot of methane on top of the liquid, they can be unhappy with alot of steam coming off the liquid.

If you strike those situations they could both be in error and still agree.
Maybe a different senseing method is called for
Just a few thoughts on my part to ponder.

 

[Binaural]

1. Most transmitters have an on/off fault signal, use this in addition to the measured level to help arbitrate between the level signals

2. Of the suggestions made, I like OkiePC's the best, with a bit of modification. I have often seen tank levels measured with an analogue probe with 1-3 digital level probes as backup for powders and crushed solids. Essentially, you check on the rising or falling edge of the digital level signal whether the level measured by the analogue sensor is valid, and you get additional robustness the more probes you have if you program them with a bit of low cunning.

3. It's worth noting that in many applications the digital point measurements are the ones used for control purposes rather than the analogue level, which is used for process visualization and tuning instead.

 

[kwires]

Just my 2 cents worth but why don't you just use 1 analog meter and a high level switch and a low level switch. Better than redundancy for this application IMHO.

 

[Tom Jenkins]

I agree with kwires - this is how it is usually done. The level transmitter is the primary device and a low-low level switch is used for all pumps off and a high-high level switch is used for all pumps on and alarming. The level switches are usually float type, and there are a lot of models available in corrosion resistant metal or polymers. Many float switches have dual contacts so you can use one for the PLC and the other can go to an alarm horn or auto-dialer etc.

If you have two analog devices the usual technique is to compare the readings and calculate the difference. If this difference exceeds a certain percentage of reading then an alarm trips. It is up to the operator to designate which probe is correct and to set the other one offline so it's readings are ignored. However, most often you still use float switches at the extreme points.

I've never been convinced that redundant analog sensors were worth the effort!

 

[Norml]

I did something similar in a wastewater pumping station where "Instrument Health" was monitored by examining bit status in the analog input modules. This was a SLC500 setup and the status bits indicated over range, under range and open circuit conditions. (Don't have the project here and I can't remember which bits were involved.) This was combined with a loss of echo alarm from the ultrasonic transducers to establish an alarm condition. We even went as far as alternating the level transducers on a daily basis to ensure that the backup was always ready to go.