Can you measure moisture levels to <1% in an air stream?

[TimB100]

I hope you can help

I would like to know if it's possible to measure the moisture content of an hot air stream that contains moisture. I'm looking for 1% accuracy.

I was thinking of say a PTFE tube wrapped with wire and an oscillating circuit set up. Then as the air passes the capacitance changes and the freq then I can measure that to determine the moisture level.

Is that possible? Any other ideas.

I'm only just looking at this so any ideas are welcome.

One last one what about charging the water as it passes. The higher the moisture level the greater the charge it will carry and hence affect something I can measure. Is that no go?

Cheers

Tim

[Kleinstein]

I don't think the change is dielectric constant will with that large with humidity, so it is only a rather small effect on the capacitance. So I would more expect a kind of highly stable air type capacitor (more like ceramic isolator, not PTFE). There should be other options to measure moisture.

It also depends whether one want relative humidity or absolute water contend. Also other gases might have a cross sensitivity.

An other method could be optical absorption in the NIR range. Either by weakening of a beam or as a photo acoustic cell.

There are also direct moisture sensors - at elevated temperature they might be reasonable fast. Otherwise the sensors tend to be slow responding as it takes quite some time for the humidity to get into the sensor.

1% uncertainty is not that easy and likely needs checks / calibration from time to time.

[TimB100]

Thanks

Doing my home work on it an LC system would be the best option. But like you said I doubt it would be sensitive enough.

My temperature range is 120-160oC and 0 to say 15% humidity.

[JimRemington]

Research brief on accurate measurement of trace water vapor (nontrivial undertaking!): http://photonicssociety.org/newsletters/oct00/trace.htm

[DaJMasta]

No idea if it's viable, but could you divert a potion of the air to sample and then cool it down and use more conventional techniques?  If it's only going to 15%, then you could cool things slightly to be in the range of commercial sensors and may still be above the dew point, and if not, you may be able to make it condense by chilling it and measure the output and have a temp sensor for the stream and do some math to get your water content.

[JS]

  What do you do know about the air stream and what do you need to know?

  Maybe measuring temp, MAF, speed and pressure (maybe not all needed) you can do the math to work out the RH. If you do need or already know some of the factors you are half way there. As it's a non direct measurement (as it will probably be anyway) you need to take into account all the error sources and add them properly.

JS

[Smokey]

I'm pretty sure any moisture sensor should work.  Most airstreams are pretty roomy....

[LaserSteve]

Talk to the folks at co2meter.com.
They may have some insight.

Steve

[CatalinaWOW]

I suspect there are some unstated requirements that make this non-trivial.  Things like time constraints on the measurement (making it ridiculous deliberately - how many measurements per week do you need, and how many months can be spent waiting for the answer? At the easy end of these scales there are existing solutions).  And as hinted earlier, what things other than water are in the stream. 

Assuming that you want responses in measured in seconds or fractions thereof, and if the stream is pure, and can be assured to contain no condensate, the near - IR transmissometer approach seems pretty promising.  Calibration will be challenging.

[T3sl4co1l]

Well geez man, 15% humidity, at higher than the boiling point of water, is steamier than the air above a pot of boiling soup!  Run it through a heatsink and use a condensation or dewpoint sensor to find that out!

How close to 0% do you really need?

If you had said 1-15%, that's easy.  If you had said 0.1%-15%, that might be challenging but not impossible.  But you said 0%, no tolerance, an utter absolute: impossible to achieve, even in the vacuum of outer space!

Knowing tolerances is absolutely vital to design.

Tim

[JS]

Well geez man, 15% humidity, at higher than the boiling point of water, is steamier than the air above a pot of boiling soup!  Run it through a heatsink and use a condensation or dewpoint sensor to find that out!

How close to 0% do you really need?

If you had said 1-15%, that's easy.  If you had said 0.1%-15%, that might be challenging but not impossible.  But you said 0%, no tolerance, an utter absolute: impossible to achieve, even in the vacuum of outer space!

Knowing tolerances is absolutely vital to design.

Tim

He did said 1%, so anything under 0.5% would be considered 0% no error. At least if that spec is correct.

JS

[jmaja]

My temperature range is 120-160oC and 0 to say 15% humidity.

0-15% humidity? Is that relative humidity (at that temperature?), specific humidity, mass fraction of H2O or volume fraction of H2O?

120-160 C at what pressure? Above boiling point of water (= no relative humidity)?

Do you need to measure on line in those conditions or can you take samples?

There are may ways to measure the amount of water in air. What do you need as an output?
Here is some basic information http://www.npl.co.uk/upload/pdf/Beginner's%20guide%20to%20humidity%20measurement%20(draft%20for%20comment).pdf
And here is a nice calculator for converting humidity values (doesn't work above 150 C) http://www.humcal.com/index.php

[T3sl4co1l]

He did said 1%, so anything under 0.5% would be considered 0% no error. At least if that spec is correct.

Yeah, but is that percent of measured, percent of actual, percent of full scale, percentage point?

Unqualified, I would take that to mean 1% of actual, which still means an absolute tolerance of +/-0 at "0%" (which is impossible, so again, it's ambiguous), and a maximum tolerable error of 0.15% total possible humidity (i.e., 0.15 percentage points).

Usually, something with a wide dynamic range like this, is specified as offset plus proportional error, which is very reasonable for both how the measurement is performed, and how it's used.  If the OP could give us these two details, we could discuss actually helpful solutions, rather than pedantic discussion of tolerances. 

Tim

[TimB100]

Thanks for the input.

It seems that its very hard and hence the reason no one has done it yet.

BTW +100 deg air with moisture could be technically thought of as steam. I was hoping to not give to much away but it just hampers things. So more details.

I would like to take a sample from a steam line (say via a 0.6mm hole this will cause a pressure drop and restrict the flow) pass that through a non contact sensor eg a tube of say 10mm id and get some idea of the moisture content.
NOTE!!!!!
I do not want to cool it. COOLING is not on the table why? Because when you cool it. it becomes 100% water!!!
Sampling is real time.

Basically it comes down to this.

Would an LC set up with say wire wound around the tube be detuned by the moisture level in the inside of the tube.
Or how about 2 plates one on an outer tube and one on an inner and the steam pass'es through the cap between them. Cap could be say 1mm.

What do you think? Could it work?

Ta

Tim

[HackedFridgeMagnet]

I do not want to cool it. COOLING is not on the table why? Because when you cool it. it becomes 100% water!!!

If its steam then its 100% water so no need to measure it.
If it's steam and air, then when you cool it, it isn't 100% water. </pendantic_mode>

[T3sl4co1l]

If it's 100% steam, and cooling it would make "100%" water*, then your meter is done: tie a wire to +V because it's always 100%!

(*The volume or pressure will change at the same time, so if you think of the container as being 'less full' because it's under less pressure, that wouldn't be "100%", in a certain sense (an incorrect -- but intuitive -- sense).  Hence the quotes, and this note.  Another example: air compressors are rated in CFM, which might be ambiguous; what they mean is CFM at atmospheric.  There's information hidden in the definition.)

It sounds like, a dewpoint sensor will determine what you need, give or take a conversion table.  If it's saturated steam, then you simply get the condensation point at system pressure; if a carrier gas is present, then the temperature will be lower, corresponding to the condensation temperature and partial pressure of the steam content.

Finding one may be difficult.  If you aren't prepared to spend a cool $1k on an industrial sensor, well... maybe you just don't need one that bad?

Ah, and that's another key piece of information: how soon do you need it, and how much are you willing to pay?

Two references:
1. http://us.magnetrol.com/Literature/1/57-239.0%20Eclipse%20Steam%20White%20Paper.pdf
This looks helpful.  Measures steam, electromagnetically, it seems.  Pretty awesome!  Bet it's like $10k, too...

Note that the dielectric constant increases very, very slightly.  You'd be measuring parts per million to get desired accuracy.  But the apparatus (likely being made of common metals like steel) will expand on the order of tens ppm, due to temperature fluctuations: so, your proposed solution would be a better thermometer than gas sensor.

It's also not specific, because the dielectric constant of steam (at a given pressure) isn't much different from that of any other gas (at the same pressure).

So we could go back and forth, picking and critiquing approaches, but you'd just end up frustrated, and blame us for the failure to come up with something.  Which is true, we weren't able to come up with anything.  Hence why I'm writing this...

2. Regarding threads like this...
http://xyproblem.info/
Consider the situation we have here.  You come to us, providing very little information about your intended application (and perhaps, personally having little knowledge about the subject -- this might be some random thing that's been thrust upon you; we have no idea!).

And you propose solutions to us.  (Well heck, if the problem is solved, what are you coming here for? )  But it's clear you aren't an expert in that field, so, I hope you aren't getting too fixed on a particular solution or anything!  Just a basic idea to get things rolling, right?

But since you're not an expert, for us to discuss those solutions, we need to convey quite a lot of information to you, about a particular approach may or may not work, so that you can make an informed decision about what is best for your application.

So by hiding your background information, you leave us to speculate wildly, and the discussion really just goes nowhere.

Since, frankly, we're a bunch of lazy engineers, who have better things to do than teach the intricacies of sensor design.  Sorry.  (It's even more likely, most of us are not sensor design experts, either!  Sensor design is tricky business, and, accordingly, commands a pretty penny for services rendered...)

And, yeah, this may be a proprietary situation, that you can't tell us much about -- but in that case, you should be searching all the more earnestly, either to increase your knowledge of the subject, or to find someone who can solve your problems under suitable controls -- such as a contract with NDA, stated work, and compensation.  These things aren't free, after all!  (And that's why buying a cool $1k industrial sensor is, very often, a darned good idea!)

Tim

[LaserSteve]

Wait till he sees what optical  resonant cavity ringdown spectroscopy costs. In the US, at my major university, a 200 litre  tank of  dry, reference Air costs 225$.  This is a pay to play field.

Steve

[LukeW]

Mid IR (thermal, 10um ish spectrum) transmittance measurement, assuming water is the dominant IR-absorbing component in the gas stream and there are no other abundant heteronuclear gases like CO₂?

Some sort of quadrupole resonance or NMR style technique, assuming H₂O is the dominant hydrogen source in the gas stream?

Neutron scattering?

[ogoun]

if the carrier gas (?air in your case) is rf transparent at 2450MHz, maybe you could use a microwave source and cavity and look at the Q change as the water vapour gas passes through it... using 2450 MHz should be effective, as water molecules should absorb energy at this frequency. Probably still need to temp compensate, though.

l8r
o

[_Wim_]

BTW +100 deg air with moisture could be technically thought of as steam.

I have to disagree with this. This is a common misconception, but theoretically steam does not contain any air. In practice however, steam does contain some air (typically referred to as non-condensable gasses), and they originate from the absorbed air in the water that was used to generate the steam. The amount in good quality steam is however very low: EN 285 states max 3.5%, this is ml of GAS vs ml of condensate. Taking into account that 1 liter of condensate is equal to approx. 1860 liter of steam, this results in 0.002% volume/volume gas. Taking this last number into account, hot air with some moisture content cannot be thought of as steam...

[nfmax]

I would like to take a sample from a steam line (say via a 0.6mm hole this will cause a pressure drop and restrict the flow) pass that through a non contact sensor eg a tube of say 10mm id and get some idea of the moisture content.
NOTE!!!!!
I do not want to cool it. COOLING is not on the table why? Because when you cool it. it becomes 100% water!!!

If your sample flow undergoes a pressure drop then it will cool anyway

[T3sl4co1l]

if the carrier gas (?air in your case) is rf transparent at 2450MHz, maybe you could use a microwave source and cavity and look at the Q change as the water vapour gas passes through it... using 2450 MHz should be effective, as water molecules should absorb energy at this frequency. Probably still need to temp compensate, though.

l8r
o

Optical absorption would be a very effective method, especially for lower concentrations of water.  But not at microwaves -- even at ~30GHz (where the atmosphere is fairly opaque), it still takes some ~km of water vapor (so, presumably, at the concentrations we're talking here, ~m still) to have a reasonable absorption amount.  But that's okay, we can just crank up the frequency!



Far IR would be extremely effective, at 3.0-3.7 cm^-1 (2.7-3.3um).  Unfortunately, detectors for this wavelength aren't cheap, and probably have to be kept cool (so you aren't going to put them directly in the gas stream or anything!).  That said, $50 for a sensor device, plus some kit around it, would give excellent results.  You'll still want to check its performance, to make sure that it responds to temperature, pressure, concentration and impurities (especially "greenhouse" gasses like CO2, CH4, etc.) in a useful manner, and to compensate for these errors.

Tim

[TimB100]

T3sl4co1l

"Note that the dielectric constant increases very, very slightly.  You'd be measuring parts per million to get desired accuracy.  But the apparatus (likely being made of common metals like steel) will expand on the order of tens ppm, due to temperature fluctuations: so, your proposed solution would be a better thermometer than gas sensor."

Thanks, just the info I did not know and was looking for!

"But it's clear you aren't an expert in that field, so, I hope you aren't getting too fixed on a particular solution or anything!  Just a basic idea to get things rolling, right?"

Yes you could say I'm an expert in some aspects of the process. I if you have not noticed I'm not explaining exactly what the application is as I have very big corporate rivals. I know the products they have developed, I know the prices, I know the market. I know just about every way of doing the job, and the patents held that I have to work around. So I have to try and to give just enough info to get the answers but not show up in a simple search about the product.

I'm automating a process of a manual system. I sell the manual system and am probably a world leader in sales in my field. I'm thought not brilliant in electronics. I can get by in the digital world and have products on sale where I designed and programmed everything from the ground up. But know a little more than nothing about analog.

The reason for this thread is that it is to check if there is an alternative system to another I am working on.

I really really appreciate the help everyone is giving me its invaluable!!!!

I'm going to spend some time reading all the fantastic advice you have taken the time to write up.

Again many thanks.

Tim

[Conrad Hoffman]

It's not difficult to build a chilled mirror dewpoint sensor with good accuracy, nor is it even expensive, but getting an overall accuracy below 1% will take a lot of attention to detail. They also get dirty- how clean is the airstream?

[CatalinaWOW]

Water vapor has a number of strong absorbtion lines between 0.6 and 2 micrometer wavelength, allowing the use of relatively low cost uncooled detectors.  There will be real costs in buying appropriate narrow band optical filters, setting up stable sources, and real difficulties in sorting out losses due to vapor absorbtion from the losses due to absorbtion in any liquid water in the stream and on the chamber walls and scattering loss from droplets in the stream in addition to any absorbtion from other gases in the stream.  Since you know your process, you know what the contaminant gases are, and by looking at their absorbtion spectra you can select a water absorbtion line to minimize the effects of the contaminants.  Often people will sense in two or more bands, with some not strongly absorbed by the gas under study.  These secondary bands can be used to help measure/compensate things like scattering losses, dirt on the windows and the like.

Also, since you have a working manual process, you probably don't need any absolute accuracy from the sensor, only a comparison of the sensors output with your existing process control.  Much easier than establishing absolute answers.

Only you, or someone you bring inside your fence can properly sort all of this out, as there are many trades that depend on the details of your process and requirements.