Pump protection against dead heading 3

[eeprom]

Hello,
I am an EE, and I'm working on a control system to protect centrifugal and vortex pumps from catastrophic failures due to dead heading. I am going to use temperature as a leading indicator of failure, so I intend to install a temperature switch onto the housing of the impeller. From a control standpoint, this is very simple: it is just a thermostat. But given the huge variation in pump sizes and shapes, selecting and mounting a good switch is not so simple.

The best solution I have come up with so far is to use a thermally conductive epoxy to mount to the pump casing. But this would be a bit of a pain to install, and it may not hold up as well as a bolt. Tapping into the casing is out of the question. So I am stuck with the question, "how to mount a switch to the casing of a pump?"

Does anyone know if a switch is already made for this application?
If not, does anyone have any ideas for mounting a switch?

thanks
EE

 

[SNORGY]

I think a flow switch either in the suction or discharge line would be more cost effective and probably simpler. Then all you need to do is configure the pump run status to come on line before the low flow switch g.oes live.

 

[Yobbo]

I do agree with Snorgy. Pumps can heat up very quickly adn locally temperature hotspots may occur. These hotspots are determining the possible damage. As soon as your temperature measurement is not at the high temperature spot or doesn't react swiftly enough the pump may be damaged even before your control system will notice it. Principally it is better to detect a lack of flow through the pump as it is the actual cause of the sequence of damages.

Karel Postulart, The Netherlands
Nuon Power Generation

 

[BigInch]

an FS will also tend to protect against potential cavitation instead of allowing cavitating and maybe flashing while heating up, provided that your flow remains above minimum flow. Keeping flow above minimum flow is better, but if not, an FS is still better than a TI.

What would you be doing, if you knew that you could not fail?

 

[eeprom]

I agree that a flow switch would be better. It was my first recommendation to my client. But the slurry being pumped is exceptionally abrasive. A flow switch (like a paddle) probably wouldn't last 6 months.

Please also note that I am not trying to protect the pump; I am trying to protect personnel from catastrophic failures of the pump.

EE

 

[hydtools]

Use a pressure switch or gauge in the outlet to indicate approaching shutoff pressure since pressure is what you are concerned with. Temperature is after the fact. And you have decided to not use a flow switch.

Ted

 

[SNORGY]

You could do a non-intrusive device like a Doppler meter with a low flow signal. Otherwise a pressure switch is another good option as long as there is enough elapsed time allowed to ensure you are detecting a true deadhead situation as opposed to just a transient.

 

[eeprom]

Thank you for your comments. I am glad to have this discussion. I have done a lot of research on this, but I am not confident that I have considered all of the possible setbacks.

A doppler device would be very expensive, especially for a smaller pump. Remember, the goal is to prevent catastrophe, not to protect the pump.

The difference in dead head pressure and operating pressure is too small to detect, and the difference will vary depending on the pump and where it is operating on the pump's curve.

Temperature leads pressure. The energy of the impeller will be converted to temperature in the fluid as long as the fluid is below saturation temperature. Once the fluid reaches saturation temperature, then the pressure will start to increase.

If the pump discharge is 50 psi, then the saturation temperature would be around 300F. If the temperature switch is set to detect 180F, the pump will shut off before any pressure builds.

 

[BigInch]

I don't see how temperature is leading pressure.

What would you be doing, if you knew that you could not fail?

 

[clay87]

1. Perhaps don't use discharge pressure for your analysis.

2. Is the concern that the equipment is going to vibrate/break and hurt somebody or that the pump case is gonna explode from over-pressure? If the latter, how about a pressure safety valve?

 

[Artisi]

Have you considered monitoring power input

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[eeprom]

I have not considered power monitoring, but I have considered current monitoring. And I found that there is a surprisingly small difference between a loaded pump and a dead headed pump. So current is not a good indicator.

I've also looked at pressure discs: these were cost prohibitive.

BigInch, pressure leads leads temperature in the same way a boiler works. The fluid (water) is initially at ambient temperature at 50 PSI (for example). The energy from the pump, which is constant, adds enthalpy to the water. While the water is in liquid form, that increase in enthalpy takes the form of heat, at 1 BTU per lb water per degree F.

Once the water reaches it's saturation temperature (~305F), then the input energy starts converting the water to steam. It takes 1000 BTU per pound of water to vaporize the water. Once all the water has been converted to steam, it is now dry steam, and the input energy goes toward superheating the steam. This is where the largest pressure increase comes from, and this is the most critical point for pump failures.

So, according to my assessment, short of a flow switch, temperature will be the first detectable change in the direction of pump failure.

 

[bimr]

There is probably no physical method to mount a sensor that would also not create a barrier to the flow of heat. You would also be waiting for the pump housing to heat before you would be able to sense the temperature on the outside of the pump casting.

I think Atisi has made a good recommendation to monitor the power input to the pump.

 

[SNORGY]

If the difference between dead head pressure and operating pressure is too small to detect, I.e. flat characteristic, then flow or power are better measures than pressure. You would need to know the power draw at operating point as well as at shut-off; depending on efficiency of pump and motor, I suspect power draw would go down as you approached shut off. My reservations about this are that as you have wear over time in service, I don't have a good practical feel for how repeatable the power measurement is going to be. Maybe it will always be accurate and repeatable enough to be effective.

Apart from that, mabe a skin temperature RTD on the pump casing is fine. People do that with exchangers and air coolers all the time. I would think you would need to do something to compensate for ambient or suction side variations in temperature. A corresponding RTD on the suction would provide the second measurement to give you the differential.

 

[eeprom]

The internal fluid pressure will certainly be higher than the casing temperature. But (in my opinion) the temperature of the casing would be within 25 degrees F of the internal temperature. My margin of safety is about 100F. My plan was to use thermally conductive epoxy to mount the switch.

I am trying to spec this for pretty much any centrifugal or vortex pump. I don't have a very good plan for fastening the switch. The switch cannot fall off. Bolting it on would be ideal, but it would also be unlikely to happen.

I am assuming that no one in this forum is aware of an existing temperature switch for this intended use. But if someone has any ideas on how to securely fasten this, I would like to know.

 

[eeprom]

SNORGY,
Why would the differential temperature be valuable?

 

[clay87]

To answer your specific question, a couple suggestions:
* pumps have ports intended for pressure gauges, vents, drains, that are often not needed by the operator and are just plugged. Maybe you could take advantage of these (and measure fluid temperature directly) instead of adhesive on the exterior of the casing?

* Maybe mount this device using exiting fasteners for pipe flanges or seal housings? Or does it need to be more strategically placed?

More thoughts:
- A 100 degree difference between the rotor and casing may lead to pump failure regardless of how close the fluid is to its boiling point. Your concern appears to be over-pressurization, so why not a safety valve?
- Goulds has an I-alert device that measure pump (bearing) housing temperature. Not sure how they mount it.

 

[eeprom]

clay87,
Thank you for your input. The use of a thermal well is problematic because the internal fluid is a very abrasive fluid. The port won't last.

I will investigate the Goulds I-alert.

Thanks

 

[SNORGY]

eeprom,

My rationale is that if you start with say 10 C water and get a 30 C rise in temperature, that might signify the development of deadhead whereas not so if you start with 25 C water. Thus, if you set the temperature switch at 30 C (one fixed temperature) you don't necessarily have a good reference point from which to establish how much heat is being transferred to the water as a result of the pump work input.

I am not familiar enough with your system to determine how much heat or temperature rise would be bad, but it is more the rise in temperature across the pump that suggests deadhead than simply the discharge temperature by itself.

 

[BigInch]

eeprom,

Thanks for the thermodynamics lesson. I had no idea that you could boil water without seeing a temperature rise first, at least without first reducing the pressure below vapor pressure isothermally. It doesn't sound like you're doing that.

Here's the pump lesson for you. If everything is as you say, you are very near dead head pressure before, or at the same time, or just as you begin generating heat in the fluid. That's already below minimum flow. That is the time to shut down, not when you have been boiling fluid for 15 minutes, heating the bearings and burning the seals. Unless you are running normally at nearly boiling temperature, before you change it to vapor, you will see a temperature increase well before you get to boiling temperature. Surely you can't mean that you boil the fluid before seeing a temperature rise! As you say, you've got to heat the fluid to boiling without experiencing a temperature rise.

Max Power and current monitoring won't help, as power and current are probably their lowest at dead head.

If you don't mind running dead head for a few minutes while things get hot and pressure builds up a little higher and it sounds like you don't, at least until it's ready to explode. Put on a relief valve set a couple of percent higher than dead head pressure, or equal, or 1/2 percent lower. Recirculate the relief discharge back to the tank. Put the flow switch on the relief line.

Now. Sounds like somebody picked a bad pump for your application, or is running it at a bad flowrate. If you're running so close to dead head pressure that you can't tell the difference from when it's dead heading or not, your normal flowrate is not anywhere near the BEP for a conventional pump, or you'd have to have one hell-of-a-flat pump curve. Or is it one of those rising pressure to BEP pumps. You should have some room there for a relief/recirculation valve, at least 10% or something. How is it that you're normally running so close to dead head that you can't tell the difference.

Your pump doesn't have a relief valve in it now? That' another way to prevent an explosion BTW.

Can't get a TI on the bearings?

What would you be doing, if you knew that you could not fail?