Meaure water temperature inside a sealed pipe

[kybert]

Hi,

I need to measure the temperature of water flowing inside a 15mm coper pipe.

The pipe is a mains pressure water pipe (like the one that supplies water to your sink).

I cant simply insert a temperature sensor into the pipe because i cannot get the wires back out. Once the tap is turned off, the hole in pipe will e pressurized and probably leak!

Can anyone think of a way of doing this? I thought about measuring the copper temperature, but i dont think the copper will heat/cool quick enough to give accurate results.

I have a flow sensor in the pipe and basicly i need to know the temperature of the dispenced water, at 1 second intervals, accurate to 1°C

The maximum pressure in the pipe is 4bar.

Joe

 

[bryan1]

Hi Joe,
Eh mate the easiest way would be put in a T-piece and mount the temp sensor off the branch, it won't be hard to seal it as 4 bar isn't really that high a pressure. The way I would do it is solder a 1/4"bsp nipple into the branch then seal a LM335 or ds1920 into the fitting. There are plenty of compression fittings on the market and that would make for a dead easy seal.

Hope this Helps Bryan

 

[kybert]

I was thinking of epoxying a DS1920 (1wire) device into the end of a push fit end stop.

Since they are made of plastic and can be fitted to copper pipes, they would be easy to make and easy to fit.

My only worry is the bond comming lose and water going everywhere.




I think i need to sit down and have a think about this.....

Joe

 

[stevez]

A couple of things -

Unless the fluid, pipe and surrounding air are all at the same temperature there will always be some kind of temperature gradient. If the fluid is hot, relative to surroundings the very center will likely be hottest, the fluid touching the pipe wall will be coolest, the outside of the pipe wall will be cooler than that - and so on. Where this might play out is if you suddenly turned on some hot water and it flowed thru cold pipes the water would be cooled and the pipe heated. If you were to look at a cross section you'd see the differences.

A reasonable question might be to ask where in this cross section is of greatest interest - the center, the outside, etc? Don't forget that the fluid might be cooling as it travels down the pipe, especially if the initial temp just went up and the pipe hasn't warmed all the way along it's length.

The whole point of this is to illustrate why measuring the pipe wall temp might be sufficient. Now and then very low thermal mass sensors are placed in fluid streams but it's more common to have a T fitting and a well - more or less a protrusion into the fluid stream with a hole in it where the sensor resides.

If your sensor is small enough you might fabricate your own well. Thin wall copper or brass tubing could be inserted thru the pipe wall and soldered in place. The sensor could be packed into it with some thermal grease. That might give you somewhat faster response but also the separation you desire from the fluid stream.

If the pipe has a rather thick wall (castings of some faucets do) you might fashion a well from threaded brass rod (small dia). You'd drill/tap the thick wall. You'd drill out a pocket in the brass rod for a sensor.

Hope this is of some help.

 

[أب&am]

thank you

 

[Oznog]

Perhaps we should back up a bit.
Do you really truly have to have a 1 sec response time?
As you note, putting an immersion probe through a round pipe wall is not the best situation. So, are you sure you don't have a block somewhere from whatever part of this heats/cools the water that you could tap threads and screw into? You want to be really close to the source of the water's temp change.

There might be some computer alogrithms which may help. For example, if you have 60F water and suddenly switch to 70F water it might take 10 sec for the pipe wall to get to 68F and longer to get to 70F. But, if you're reading 60.0F and then see the temp go to 61.0F in the next second, the high rate of change can be used to calculate the actual temp. Like a PID controller. You would need to characterize the properties of the pipe and the thermistor itself.

 

[evandude]

As others have said, getting the sensor in the pipe without leaks is really not as big a problem as you might think. a little bit of decent epoxy (I would suggest JB weld) should be far more than enough to prevent leaks, and if you are really that worried about it, when you install it in the fitting or whatever, seal around the inside as well with some silicone or something.

 

[ante]

Hi Joe,

If response time is not extremely critical the copper tube is not a problem here. Just attach the sensor to the outside of the well-cleaned tube with some thermal grease on it and some isolation material around it all and it will work fine.