Thermistor sensor on outside of pipe to test fluid temp

[fastline12]

We are designing a temp sensor to test fluid temp of a coolant. Problem is the coolant cannot be easily drained to insert a probe in the coolant stream. The two options we have are to design a puncture or penetration probe to insert in the rubber hosing that transfers the fluid or to simply attach the probe to the outside of the hose and build a custom calibration for the sensor to read on the outside of the rubber hose.

I am kind of concerned as to the thermal reactivity of the outside of the rubber hose.

Just looking for some experience here to decide which direction we should go. Common sense says get into the coolant stream and keep things very accurate but that could prove problematic.

 

[MintJulep]

"thermal reactivity"

That's a very fancy technical sounding phrase.

What's it mean?

 

[fastline12]

oops, lets put "thermal conductivity" there. I am concerned that this value may vary from hose to hose. We have a design accuracy goal of +/- 1*F at 200*F. If this method is being used in the industry elsewhere, that might help our decision I guess.

 

[ione]

Try this

 

[fastline12]

Thanks for the link. It was not clear but appeared to be referencing metallic pipes where thermal conductivity would be higher. I am not sure if reading through a rubber hose would even be worth the testing.

 

[rb1957]

1) how steady is the coolant temperature and flow rate ? i think you need both to be steady in order to have a uniform heat flow into the pipe.

2) you'll need to measure the ambient temperature as well, as this too will affect the temperature of the pipe.

3) is this a one off installation, or will you be meauring many installations (ie many pipes, all the same part number) ? i'm thinking about the difference between measuring a single pipe or different pipes with the manufacturing variabliity to be considered.

4) plus/minus 1degF sounds awfully tight to me ...

 

[ione]

Ok, my miss, that stuff was definitely for metal pipe (high conductivity).

You can proceed this way (assuming steady state).

Qout is the heat (per unit of length) due to convection from the outside of your rubber pipe to the ambient (neglect radiation as probably you are not working with high temperatures).

Qout = 2PI*rout*hout (Twout – Tamb)

Twout = pipe outer surface temperature (you can measure it)
Tamb = ambient temperature in the surrounding of the rubber pipe (you can measure it)
hout = convective heat transfer coefficient (assume it is 5 W/(m^2*K) for natural convection)
rout = outer pipe radius

Qcond is the heat (per unit of length) transmitted with conductive mechanism through the pipe wall

Qcond= 2PI*lambda*(Twin – Twout)/ln(rout/rin)

Twin = pipe inner surface temperature (unknown)
lambda = rubber thermal conductivity (you have to know this)
rin = inner pipe radius

Qin is the heat (per unit of length) due to convection from the inside of your rubber pipe to the fluid.

Qin = 2PI*rin*hin (Twin – Tf)

Twf = fluid temperature (unknown)
hin = convective heat transfer coefficient (it depends on your fluid, its flow regime and its temperature)

In steady state it must be

Qout = Qcond = Qin

From: Qout = Qcond you can find Twin

From: Qcond = Qin you calculate Tf and then check back whether hin you’ve assumed converge or not.

It is a trial approach and it will give you an idea of the fluid temperature.

 

[rb1957]

a thought ... aren't there plenty of natural breaks in the pipe where you can insert a spacer with a thermocouple, so you can detect the fluid temperature directly, rather than puncturing the tube wall.

is the pipe lagged ? i'm guessing it is ... this might insulate a probe on the outside of the tube from the surrounding environment.

personally, i'd test the installation. if you can't test the pipe you're measuring (maybe it's installed and in process) then build a test piece, run with fluids of known temp.

 

[fastline12]

There are some breaks in the pipe where a sensor could install but most all would require the fluid be drained and this is also a "universal" design to be used with nearly any rubber hose.

In some discussion here, I am pretty concerned of the delta Ts we will see in the piping. The piping OD is subject to ambient outdoor temps which can fluctuate radically as well as the fluid temp that will also not remain constant.

In short, I am kind of thinking testing on the OD of a rubber pipe is a bad idea in this application. Just thinking out loud here.

 

[MiketheEngineer]

Maybe you could build a test rig with an internal and external sensor. Then run it up and see if you can find any correlation??

 

[rb1957]

i think you can deal with the ambient temp with a separate thermocouple, but "one size fits all" sounds unlikely that you'll get the accuracy you're looking for ... different tube materials having different thermal prop's ?

 

[Compositepro]

It is simply a question of insulation and what response time you need. Although rubber is not a good conductor of heat, the outside of the tube will be at the same temperature as the inside, if you insulate it well. Response to temperature changes will be slow.

 

[zekeman]

How about breaking the rubber pipe and inserting a metal section between the breakpoints; and then you could mou nt the thermister to the surface.

 

[MintJulep]

Sounds like a total non-starter to me.

You could modify ione's approach a bit to include a layer of insulation.

But you'd still need to know the hose thermal conductivity and still need to make assumptions about heat transfer coefficients at various surfaces.

You want this to be universal, but hoses are not universally of the same construction, and hence not the same conductivity.

With so many unknowns and assumptions built in you'll never hit your accuracy goal.

 

[fastline12]

Agreed. After some discussion and info here, I think we will have to further research ways to get a probe in the fluid path.

 

[gruntguru]

Try posting in the Heat Transfer and Thermodynamics forum.

What is the pressure in the system? Is the rubber hose attached with hose clamps? It is sometimes possible to loosen a clamp and slide a bead-type thermocouple (lubricated with silicone sealant) between the metal and rubber to get to the coolant.

Engineering is the art of creating things you need, from things you can get.