Humidity measurements in compressed air

[BobL]

Warning - Larger Volume of Nerdy science/engineering content.

I’m posting this here because it's more of a measurement/electonics project than metal work and may interest a (very) small group of like minded folks.
My health and sleep have been poor lately so I'm restricted in what I can do anywhere including in the shed, Typically I do something for 1/2 an hour in the shed and then have to rest for about the same length of time so I have been undertaking some soft distracting projects and things just take a lot longer than usual.

Recently I started upgrading my 8 year old 1/2” galv pipe reticulated compressed air system.

The potential flow restrictions starting at the compressor where the 1/2” BSP tank outlet was connected to the retic line via a 1/4” ID 4 way, brass junction, and several short (300mm) long segments of 1/4” ID SS braided gas hose in line with the 1/2” galv pipe, all have ben replaced with 1/2” BSP fittings and hoses.

About half of the 20 odd 1/4” outlets and quick release hose Nikko style fittings inside the shed that were corroded/damaged/leaking have being replaced with genuine 1/2” Nitto, “One Touch" fittings.
A 1/4” hose and a damaged 3/8” hose inside the shed have been replaced with 1/2” hose so I have a couple of hoses with good flow.

The water/corrosion issue had been mostly eliminated a few years back when I installed an auto vent valve that opens the tank drain under the tank for 1/2s every 45 minutes - I leave my compressor and auto vent valve on most of the time so the vent valve will release any water condensed inside the tank. The effect of the vent valve is easily observable from the amount of water released from the reticulated piping system "low point" where another drain/tap can be opened to release water out from that point. Depending on compressor use and how often that drain was opened, prior to the installation of the auto tank vent valve the amount of water coming from the low point drain would be as much 100 mL whereas after the auto tank vent valve was installed it was rarely more than a few mL. However, when I run the compressor continuously for more than a few minutes the compressor tank does not get a chance to cool down and condense the water and so the auto tank vent valve has very little effect. I had a small inline water trap/filter at the end of one of the piping arms inside the shed and could easily see water appearing in the sight glass. I thought about installing that in the main trunk line but the small size of the inlets and outlets of the small water trap/filter would almost certainly stymie the air flow so what I wanted was something larger.

While i was thinking about this, literally at my feet was a box of assorted plumbing stuff I had scavenged from a skip at work many moons ago, and the box contained a standard CUNO SS water filter, and it dawned on me that a few $ worth of brass plumbing fittings would enabled me to install it into the compressed air trunk line as shown below.
377820
Instead of hard plumbing the filter/water trap inline I decided to instal it using 1/2” compressor hose and 1/2” BSP and Nitto One Touch fittings so it can be easily installed and removed.
All up the 4 quick release fittings are not cheap (~$60) but I figured if it all fails I can recover and reuse these fittings elsewhere in the system.

I’m figuring the CUNO will initially will act more as a water trap than a filter but I though I would insert a standard $4 , 0.5 micron water sediment filter, which I already have a stash of for use on my under sink water filtration system for my coffee machine, into the CUNO. The sediment filter should also prevent pretty well any crap coming from the compressor to the outlets. The intention is to eventually surround the the sediment filter with desiccant to see how much water in can pull from the air.

Then I thought I would like to know how efficient any of these measures are at removing water from air so I knocked up a measuring device incorporating an in-line humidity sensor.
377821

The sensor (S) is the very widely used relative humidity and temperature sensor known as a DHT22, and is located inside the black plastic 32mm T-piece fitting (SC).
Using Nitto fittings on the sensor housing means it can be placed anywhere in line although the primary positions would be in the pre and post filter locations.

A DHT22 clone costs ~$5 and connected and read by a "~$20 all up" box of Arduino based electronics that took me (even at my slow working speed) about a day to assemble and program
For comparison/reference data I added second sensor and its located outside the shed next to the compressor intake air filters.
377822


Because the DHT22 outputs relative humidity (RH), in the case of how much water is in compressed air, rather than RH it is more useful to know the Absolute Humidity (AH) ie mass of water per unit volume of air. It turns out that if you know the temperature of the air then AH can be calculated using a couple of longish formulas which the Arduino micro controller handles easily.

The screen output shows the Saturation Water Pressure (S) in kPa and the Absolute Humidity (H) in mg/L for both of the DHT22s (I = inline sensor and O = sensor near compressor air intake outside shed)
Flipping the switch on the top of the box toggles the display to reveal the RHs and Temps for the same sensors.

The H readings (typically 6-10 mg/L) obtained for outside the shed made sense although they were at times considerably higher than that reported by the BOM - more about that later.
For the in-line sensor (I) the readings I got were somewhat puzzling (ie much higher) and after a bit of reading I found out that the in line pressure (90 - 135 psi) could screw up the operation of these sensors.
These sensors operate by measuring the change in the capacitance of a capacitor which absorbs and releases water between the capacitor plates depending on the humidity and the pressure.
It is possible to make corrections for pressure but it also appears that the high pressures can damage the sensors permanently.

All this means this sensor cannot operate reliably inline under pressure and so the compressed air has to be be somehow released into a sampling chamber without changing its humidity. The big problem is that decompressing air cools it and as it mixes with air from the atmosphere this all change its humidity.

The "proper"way to assess the dryness of compressed air is to measure its "Dew Point" ie the temperature at which the water in air starts to condense, the lower the temperature required the less water there is in the air. Half decent Dew point instruments are not cheap and even then the most accurate measurements are not usually done inline but by bleeding a small amount of air (~1 l/min) from a compressed air line using a needle valve and passing that air through a thin warming Cu coil into a small chamber which is effectively at atmospheric pressure but has not mixed with air from the atmosphere.

I figure if this approach works for dew point it might work for the DHT22 so this is what I am in the middle of setting up. I will post more of this as I build this system up. With the assistance of a mate, except the olives for the coil compression fittings, I have cobbled together all of the parts at minimal cost. It may well all come to naught but it's a low key project that keeps me suitably entertained while I feel as I do.

Back to the environment of the compressor located in that enclosure in a ~1m wide gap in between the shed and the back fence. The enclosure sits on top of a 600mm high limestone retaining wall up against the shed. Also In that gap is a narrow garden bed with a rampant vine that I hack back using electric hedge trimmers approx every 4 weeks just so I can walk through that gap. The vine grows about 1m above the fence line providing good screening against the difficult neighbour on that side otherwise I would have removed it long ago. It's watered via regular garden retic which soaks pretty well everything in that gap including the limestone retaining wall and because its all so confined there's not much ventilation so it's a more humid microclimate environment than elsewhere in the back yard. This probably explains the higher than usual humidity levels I'm seeing in outside air especially on watering days. To cool the compressor when its running, a small 240V fan in the false ceiling of the enclosure draws cool air from outside the enclosure through a 5mm wide gap under the bottom of the enclosure doors, and pulls the warm air out of the enclosure through the false ceiling. This means the fan is constantly sucking moist air into the enclosure, so all in all not that good an air intake setup for a compressor to begin with.

However, I thought I might close the gap underneath the doors and install a 100mm PVC snorkel such that the fan will suck drier air from up near the shed roof through the snorkel and into the enclosure. As I already have the PVC and Junctions to do this its not going to cost me more than a bit of time. Even using the current sensor setup I can assess if its worth doing.

Sorry about the long post. Hope it was not too boring.

[lamestllama]

BobL

Interested to see where you go with this.

Eric

[benhoskin]

Bob, I built with Terry a few years back a similar setup. Except is was measuring the humidity in the gas flow at 3000 psi (207 bar) and 82 deg C. I ran into the same problem of the capacities sensor's characteristics changing due to the 200 x pressure increase that it was designed to operate at. We operated at the full P and T, just had to come up with our own calibration curve with know humidity at P and T steps. On speaking with the swiss manufacturer of the sensor they were impressed that the sensor operated at those conditions and wanted our calibrations.

[BobL]

Thanks Ben.
Yeah sometimes you just have to try things to see if they work.
Don't happen to have a spare sensor laying around :D ?

[benhoskin]

I should have, let me check on Tuesday. Glad to see someone else trying to measure the water content of compressed gasses while still at P

[BobL]

I should have, let me check on Tuesday. Glad to see someone else trying to measure the water content of compressed gasses while still at P

Wow! - I was kidding, but it would be interesting to have a crack at it.

[OxxAndBert]

Interesting project Bob. Thanks for posting and I hope your health is on the improve.

Steve

[BobL]

Thanks Guys,

Despite my ongoing health and sleep issues I was able to knock up a couple of olives this afternoon and assemble and quickly test and almost complete absolute humidity measuring system for compressed air.

The olives are about 6mm long and have a max OD of 6.4mm, and while they are a tad on the rough side but I only had ~20mm of stock to work with so not much room to manoeuvre.

Those on the left (~3.5mm ID) are the ones I already had, whereas mine are the slightly longer pair on the right and fit my 1/8" Cu tubing.
377827
It turned out that a 1/8" drill bit produced a bore in the olive with a very nice tube fit.
I then annealed the the olives in a flame and tightened the cap nuts up lightly, then with a spanner until the sound of escaping comp air was below my hearing threshold and then used the detergent and water leak bubble test and about a 1/2 turn was enough to stop the bubbles.

So here below is the almost complete test cell setup.
377828

Air is currently sampled from either the filter inlet (I) or outlet (O) by switching the braided SS hose to I or O.
Next week I will purchase some "quick fit" 1/4" plastic pressure tubing fittings and permanently set up a T-piece and pair of small plastic taps in place of the braided hose. Switching between I and O will then just involve switching the plastic taps.

The sampled air flow is controlled by needle valve 1 (NV1) which leaks a small amount of air (~1L/min) into the ambient temp coil (ATC) which expands and warms the air to ambient T before entering the black plastic-T sensor chamber where the HDT22 sensor makes continuous readings every 2 seconds. Preliminary testing shows the readings take about 1-2 minutes to stabilise. This is because the fresh air has to flush out the old air.

When NV1 is fully open Needle Valve 2 (NV2) can be used control the flow and allows for humidity measurements to be made under compressed air pressure. A calibration curve could be then constructed for a sensor that is not damaged by high pressures.

Still to add is a small gas flow meter so that consistent escape flows can be established. The small plastic taps mentioned above will be used to isolate the test cell when it is not being used - ie not using the compressor so the needle valves can be left in the same position.

Close up of just the test cell/chamber
377829
Apart from the black plastic fittings and the sensor, all other fittings, valves, the hose, coil etc for the test cell have been scrounged so it has been a low cost project.
The most expensive single items are going to be those little 1/4" plastic taps I mentioned above so the gas can be sampled alternately the I and O sides of the filter.
It would have been cheaper but way more clunky to use 1/2" BSP brass taps.

With a flow rate of 1L/min this setup only uses 60L/hr which is not a lot but I will only operate it when I want to check the moisture levels to see if the filter needs attention.

I have not taken a lot of measurements but below (I) is sample reading from the outlet of the filter 3mg/L and O was the current reading inside the compressor enclosure 8.3 mg/L.
It is reassuring that the moisture content of air in the lines is lower than the outside air.
This is probably just the effect of the compressor tank auto vent valve as I don't expect the large filter to contribute much at this stage. I also expect the results to improve as the insides of the test cell dry out. A SS test cell would have been better and I may make that one day, I just wanted to test out if it works..

377830

More testing over the next week or so.

[BobL]

WARNING: More nerdy stuff.

I have been thinking some more about this and would like to add some comments/info about the following.

1) Calibration under pressure

As I said above the twin needle valve setup I show above enables measurements to be done under pressure. This (as Benhoskin states) assumes the sensor will operate under pressure as it may not.

There are two possibilities for measurement under pressure,
a) sample the compressed air by venting a small amount to atmosphere but measure on the pressure side of a fully open NV1, and control flow with NV2,
and
b) close the sample loop and return the gas to the pressure side.

So, if you can do this measurement at atmosphere why would you bother with measurements under pressure? There are many reasons for doing this including the value and toxicity of the gas, or as maybe in Ben's case the environmental measurement conditions are restrictive. None of these really apply to compressed air but if the measurements were performed under pressure it would be possible to eventually not lose the sample compressed air from the system ie make it a close loop system.

To do this the sensor has to first be calibrated under pressure, so the first thing to do is make a series of Absolute Humidity (AH) measurements at the same compressed air line pressure. ie measure AH at atmosphere and then AH under pressure. These are recorded along with the compressed air line pressure. The air line pressure is then systemically changed (ie start at full compressor pressure and release some air from the tank) and further calibration readings taken. For a compressor with a switching range of say 90 to 120 psi you might calibrate from 90 to 120 psi in 5 psi steps. From these readings, calibration function/curves or lookup tables showing "atmospheric AH" values for given "pressurised AH readings" at a specific air line pressure can be generated.

Then all subsequent readings can be performed under pressure and the function can be programmed direct into a micro controller and pressurised readings corrected and displayed. To do this automatically would require using a pressure sensor hooked up to the microcontroller. Such sensors can be had for ~$10 and I have ordered a couple to have a play with.

A much simpler but laborious way would be to take any pressurised AH reading and lookup the corresponding atmospheric AH values for a given air line pressure from the graph or table described above. The latter will not be very fine grained but they might not need to be especially give the humidity readout tolerance of the HDT22 is ~+/-2%.

Then the small volume of sampled air under pressure can be draw from a T-piece in the line, run past the sensor and return it to the main line. there are a few tricks needed to get the air flow happening properly but it's not that hard to do. The sensor could then be left on all the time and not lose any compressed. Whether I go this way or not will depend on how bored I get not being able to do much in the coming months.

2) Measurements using a consumer level weather station style sensor (ie no micro controller)

What about if you don't want to or unable to deal with micro controllers? There are plenty of inexpensive consumer level wether stations around that measure relative humidity and temperature that might be able to be used. You would need to be able to maybe removed the sensor from the weather station so that the sensor can be encloses inside the sampling chamber and in the first instance operate the sample chamber only at atmosphere. Assuming this is all possible, how to you convert an RH reading into to AH.

Well here is a conversion graph and table you could use.
377833

377835

The way to use the graph is to select the closest RH curve eg 50% (pale blue) and then proceed along that curve until you are above the temperature the RH reading was taken (eg 25ºC) and then scoot across to the Y axis and and read about 10 mg/L - the table says 11.5 mg/L - I can provide the equations or finer grained charts/tables/graphs if you need them.
Remember this is for atmospheric pressure readings. If you want to do this under pressure calibrating it will be a PITA but it is possible but you will need a number of conversion charts/tables/graphs to cover all the air line pressures.

3) Wifi/remote connects
If you decide to use multiple sensors and want to avoid cables between theHDT sensors and the micro controller from cluttering up your workshop then the HDT22's + a mini-micro controller plus a wifi module are cheap, play nicely together and are not that hard to setup so that several could be used feeding back to a master micro controller. The units will still need power so being close to a GPO will be handy.

Something else to waste a shed lot of time on when I get bored, ie unlikely.

[BobL]

Back to the environment of the compressor located in an enclosure with a small 240V fan in the false ceiling of the enclosure draws cool air from outside the enclosure through a 5mm wide gap under the bottom of the enclosure doors, and also pulls the warm air out of the enclosure through the false ceiling. This means the fan is constantly sucking most air into the chamber, so all in all not that good an air intake setup for a compressor to begin with.

However I thought I might close the gap underneath the doors and install a 100mm PVC snorkel such that the fan will suck drier air from up near the shed roof through the snorkel and into the enclosure. As I already have the PVC and Junctions to do this its not going to cost me more than a bit of time. Even using the current sensor setup I can assess if its worth doing.

This morning I did some measurements of the absolute humidity of the air on the limestone floor of the enclosure where the air enters through a gap under the enclosure door and compared those measurements to some taken at a height of ~ 1.5m above the ground outside the enclosure. The "floor air" was consistently between 20 and 25% moister than outside air suggesting a snorkel could be worth considering - I'll make a some more measurements before I decide if it's worth doing.

[BobL]

I made some more air moisture measurements during the day (which almost reached 40º) inside and outside the compressor enclosure and at compressor floor level and 1.5m above the enclosure.
Why I found was that apart from the early mornings, the air inside the compressor enclosure consistently contained more water than the air outside the enclosure. Then it dawned on me, of course it does because that's where the EFFIN compressor tank auto drain valve dumps its water. So it appears I have been repeatedly partially recycling water through the compressor with this setup. IE, compressor sucks in air, vents tank water onto the limestone floor of the container, and then water evaporates and fills up the enclosure with moist air, so that it can go through the compressor again.
DANG!
:doh::doh::doh:

What I need to do is add a short length of tubing to the auto vent drain so that it vents outside the enclosure, but then also don't draw air into the enclosure from down at ground level but through a snorkel with its external opening well above the ground. Finally I need to seal the limestone floor of the enclosure to prevent moisture wicking back inside the enclosure from rain and garden retic.

Then to see what would happen I ran the garden retic n for the usual 10 minutes waited for another 10 minutes and got the enclosure vent fan running. As to be expected the effect was quite dramatic as this raised the air moisture content inside the compressor enclosure to about 50% above what was already in the enclosure. Even at 1.5m above the ground there was easily a small but detectable increase in the MC - might have to use a taller snorkel.

[BobL]

Fixed up a few compressor enclosure things.

Added a drain hose so the compressor Auto Drain Valve (ADV) now vents outside the compressor enclosure.
The water droplets in the hose is the amount of water that stays behind after about 15 discharges.
It discharges for 0.5s every 45 minutes.
Arrow "A" shows how the air got into the enclosure - this gap has been sealed up with some foam rubber.
377890

Then I added a snorkel.

This shows the amount of space between the shed and back neighbours fence and the compressor enclosure and the rampant vine that grows up the back fence.
A 100mm 240V fan located in the false roof pulls air into the enclosure through the snorkel which is just visible above the enclosure roof.
The venting fan runs continuously when the compressor is running and for a programmable time eg 10 minutes minutes after the compressor turns off to help the compressor and motor cool off.
This compressor has no fan or air flow across the cylinders but with the new arrangement it has.
377891

View from other side of the enclosure
It's just some 100mm PVC DWV pipe offcuts and 90º bends I had laying around under the house.
The flared end is a "seconds" from the ones I make for woody dust collection.
377892

And heres what it looks like inside the enclosure the inlet comes out quite close to one of compressor filter banks.
"S" contains a DHT22 sensor permanently located inside the enclosure.
377893

Right now I've left the vent fan running to dry out the damp floor of the enclosure.

I was a bit concerned that the snorkel might release too much noise from the enclosure but it is a comparatively quiet (twin belt driven) compressor anyway.
Sound pressure level wise, with a basic urban background of 54dB (ie compressor off) the SPL with the compressor running is 56dB at the top of fence line.
Immediately in front of the flared air intake inlet above the enclosure roof it is 75dB but it rapidly subsides as the meter is moved away from the opening.

It will take some time to dry that enclosure floor out so until that happens I can't really tell if this has been worth the effort.

[BobL]

Just picked up 2.5kg of free Silica gel crystals from a mate.
377898
They're a mix of orange and dark red. The orange are indicator crystals and turn green when saturated. Guess what's going into the filter :)

[BobL]

As I said post #8 to assess how effective any filter is I need to be able to compare the AH pre and post the filter so I need a way to sample a small amounts of air pre and post the filer and divert the these to the sensor.
I ordered some 1/4" quick fit pressure connections online but decided not to wait so had a bit of a look locally and found that Blackwoods prices were almost comparable so went and picked them up this morning.

So this is what it currently looks like.
The quick fit setup enables me to easily divert either the inlet or outlet sides of the filter to the sensor.
It also means I do not need to use the needle valves to shut off the air flow.
The needle valves can can say in the same position so the I or O always see the same flow.
Those few quick fit fittings used (3 BSP adapters, two elbows, 2 taps and a T-piece) cost almost as much as the Nitto fittings
377899

By switching the taps back and forth at about 2 minute interval and watching the AH readout on the microcontroller I have not been able to tell if the water sediment filter removes any water. If it does, it is less than ~5%. Tomorrow I might try adding the desiccant.

[Pumpkinate]

I bet the desiccant makes a big difference

[BobL]

I bet the desiccant makes a big difference

yeah the bloke at Uni I got the desiccant from was using it in a similar situation - drying nitrogen for a lab application. At least I feel confident about measuring it or something, although it won't surprise me if the moisture goes down below my detection limit.

[Stustoys]

Morning Bob,

Interesting as always.

I think I see an issue you haven't addressed.(or more likely, you have thought about it and haven't mentioned it or I missed it)
Isn't the air just going to take the shortest route though the filter? As the filter is unlikely to get blocked to any great extent the air will keep taking the same short path through the very top of the filter/desiccant, this will saturate quickly leaving the rest dry?
Perhaps wrap something around the filter to force the air to travel down through more of the desiccant?

How about a check valve on the inlet or you might end up with a desiccant cannon ;)

Regards

Stu

[BobL]

Morning Bob,
Morning.


Interesting as always.
Thanks


I think I see an issue you haven't addressed.(or more likely, you have thought about it and haven't mentioned it or I missed it)
Isn't the air just going to take the shortest route though the filter? As the filter is unlikely to get blocked to any great extent the air will keep taking the same short path through the very top of the filter/desiccant, this will saturate quickly leaving the rest dry?
Perhaps wrap something around the filter to force the air to travel down through more of the desiccant?

How about a check valve on the inlet or you might end up with a desiccant cannon

All very Good points Stu and I have indeed been thinking about them.

In the first instance I will leave the spun polyester water sediment filter in place and pack desiccant around that - this will prevent any desiccant and desiccant dust from leaving the filter and entering the rest of the pipes etc. There is a small pressure difference across the water sediment filter and while this pressure difference is small it will be much greater than any caused by the desiccant so that will tend to spread the flow out - I guess I will find out how much by making some measurements.

Then I will measure a AH I/O profile over time. ie how long does it take for the O air to become the same as the I air.

Your suggestion of wrapping something around the filter to force the air to travel through more of the desiccant is one I have been contemplating and is what I will probably try next. Again measuring an AH I/O profile over time.

There are lots of other things I could try. If I made the water sediment filter slightly conical (I wonder if they can be turned?) this will reduce the pressure drop across the filter along its length so if the narrowest part of the cone is located at the furthest path this will promote flow in that direction.

Perhaps a combo of the filter wrap plus conical shape might optimise things further.

The water sediment filter does take up a long of space so ultimately I'll be looking to find something that is thinner so I can pack more desiccant into the filter holder. If I can turn the filter down I could maybe turn it into a thinner cone?

[BobL]

This morning I did a trial run with some desiccant.

As a precaution against blowing desiccant into the downstream rectic pipes I decided to leave the water sediment filter in and pack desiccant into the remaining space around the sediment filter.
As you can see there's bugger all room to fit much desiccant in around the sediment filter so the shortest pathway through the filter does not need to travel through many dessicant beads but nevertheless some 600g of beads managed to squeeze in there.
377910
The ragged looking end of the filter is because this filter is about 3 mm too long to fit inside the CUNO filter housing so I cut it to length on the bandsaw. The circular indentation (orange arrow) in the end of the filter is where the knife edge seal of the top of the filter housing bites into the sediment filter. There's even a strong likely hood of air bypassing the beads altogether because the area outside the knife edge seal line on the top/end of the sediment filter is not sealed. I will have t think about how I will seal this. After seeing all this I was not expecting too much or anything from such a short air pathway through the beads from this arrangement but though I would give it a shot anyway.

I then connected every thing back up and set up a measuring regime to measure and record the absolute humidity (mg/L) of
a) the enclosure air, b) the filter input air, and c) filter output air .

Measuring a) is easy as there is a dedicated sensor in the enclosure for that measurement.
However, because I only have one test chamber/sensor to measure b & c I have to alternate between b&c and wait for the new air to flush old out of the chamber and if the chamber has picked up moisture it needs to dry out (this is why SS is usually used as a chamber instead of plastic). These measurements require at least a minute between them so I went for 3 minute intervals to make sure.

Despite these limitation I got some useful info.

377909
The blue line shows the enclosure air - it's a warm day here in Perth, BM shows RH at 47% my enclosure meter showed 45%

Green dots show pre filter AH, (~2.5 mg/L) or about a factor 3 less than enclosure air - this water removal is the effect of the auto drain valve - it really does do the heavy lifting during light compressor use. Of course if the compressor runs continuously it's not as good as shown here - the compressed air simply does not get enough time to cool down and allow the water to condense and then drain.

The red dots are the effect of that thinning layer of Silica gel crystals around the sediment filter. As you can seen the AH was still slowly dropping dropping as all the surfaces inside the filter, the sediment filter itself and the plastic tubing would have a film of water all over them and this need to be desorbed to make proper measurements - the lowest reading after some 45 minutes was 0.6 mg/L of water. So from about 8 down to 0.5 mg/L or a 16 x reduction - I admit to being surprised.

[New problem: at these low levels of absolute humidity the DTH22 sensor is reading only ~4% RH and given its only good to +/-2% anyway, measuring anything at this level may not be that valid.]

Now these are only spot or bleed tests (~1L/min) and are unlikely to be indicative of the water removal capacity during high volume flows. To test what happens in a more realistic situation I performed 2 blast tests. ie release a 30, and the later a 60s, highish volume air releases.

The blue dots show the AH starting at 0.6 mg/L at before the 30s blast at the 20 minute mark, with readings every 30 seconds after the blast - the highest it went up to after the blast was 0.8 mg/L and then it went back down to 0.6 mg/L over 7 minutes.

The 60s blast was at the 29 minute mark and this time the AH rose to 1.1 mg/L. This time the blast triggered the compressor to recharge - note slight rise in enclosure air humidity as new air is sucked into the enclosure. it also took 12 minutes to come back to 0.6 mg/L, again surprising considering the thin layer of desiccant in use.

I though these results were quite encouraging given the piddling short air flow path way available through the desiccant

Because it is easy to do the next thing I will try is to cover up the outside top half of the sediment filter and repeat the above. This will force the air to go through a longer pathway.
Suggestions for ways of covering up would be useful. I was thinking cling wrap.


I want to optimise the desiccant layer and sort out the other filter issues before measuring the effect of some long runs of continuous compressor use.

Ultimately I will need a thinner sediment filter so can pack more desiccant into the filter holder but I think I will run out of measurement sensitivity well before I reach maximum water absorption.

[Stustoys]

I was thinking cling wrap.
Same.

I assume(what could go wrong there lol) that air flow through the filter is never going to be an issue.(I figure if you can run 40 odd litres of water a minute through it, air isn't going to be an issue at any flow rate you're likely to be talking about??) If so perhaps going forward you could cut the filter in half(or even less)and replace the top section with a pipe of some description? Heaps more desiccant over a much longer distant....
A test with cling wrap sounds easier.

The check valve was in case you have your reticulation system full and disconnect the tank side of your filter. Wouldn't be the end of the world but I smashed a 2" x 2" x 1" container of desiccate in the kitchen one time, was finding bits for months, that didnt go down so well lol. Maybe I'm over thinking things and you're less likely to do something that stupid :).

[BobL]

Same.

I assume(what could go wrong there lol) that air flow through the filter is never going to be an issue.(I figure if you can run 40 odd litres of water a minute through it, air isn't going to be an issue at any flow rate you're likely to be talking about??) If so perhaps going forward you could cut the filter in half(or even less)and replace the top section with a pipe of some description?
Yep - same wavelength here.
I have already located a piece of 50 mm ID SS tube that (~15 mm smaller than the sediment filter diameter) that will sit neatly outside the filter end knife edge seal on the inside of the filter housing.
My plan is to turn down about half the length of the sediment filter and slide a piece of the SS tube about half the length of the sediment filter onto the turned down section of the sediment filter. This will force the air to go right down into the desiccant to get around the SS tube.


Heaps more desiccant over a much longer distant....
A test with cling wrap sounds easier.
Yep will do that in the first instance


The check valve was in case you have your reticulation system full and disconnect the tank side of your filter. Wouldn't be the end of the world but I smashed a 2" x 2" x 1" container of desiccate in the kitchen one time, was finding bits for months, that didnt go down so well lol. Maybe I'm over thinking things and you're less likely to do something that stupid :).
I wouldn't bet on it.

[BobL]

SUCCESS!!!!!

Just tried the cling wrap.

1) From an outside AH of 7.5 mg/L, and an air filter inlet AH of ~2.5 mg/L, the filter outlet AH decreases about 5 times more rapidly than without using the cling wrap.
2) The minimum AH displayed is 0.2 mg/L. However, I strongly suspect it's lower than this because at the current temperature the RH display is 1% and the sensor will not read less than 1% so am basically limited by the bottom end sensitivity of the sensor.
3) Did a couple of 60s air release/blasts and the filter outlet AH does not change during the blast or thereafter - it sits on 0.2 mg/L

Thats a reduction of absolute humidity of at least 97% - I'm calling that a win!

The next thing I need to check are some longer repeated air releases to see if I can get the filter inlet AH up and then see what it can do.

[Stustoys]

:2tsup:

[BobL]

Got stuck inside with the heat yesterday so decided to whip up a new Arduino sensor micro controller box for the temp/humidity monitoring of the comp air.

377970


The old micro controller had 2 sensors ie inside and outside air.
This new one has built in wiring/sockets/ports for up to 5 sensors (S1 to S5).
The sensors inputs are set up as 3 digital for (DTH22 on S1 - S3) and 2 analogs on S4 - S5, but the sensor inputs can easily be swapped around on the Arduino Board for added flexibility

S1 currently monitors the compressor enclosure air.
S2 the pre or post sides of the compressed air filter .
S3 the air inside the shed.
My plan is to use S4-S5 for the 2 analog air pressure sensors I have ordered that will go pre or post filter.

The "knob" switches/togges between LCD display pages/screens for
Temperature
Relative Humidity
Saturated Water Vapour pressure
Absolute Humidity
Air Pressures

It turns out the DHT22 does have an extra significant digit for its RH readout but its lowest RH output reading is still 1.0% whatever its reading, ie any less than this is still shows 1.0% RH.
I have no idea how meaningful the second digit is but assuming it has some value I decided to display and play with it.

The smallest AH reading that can be calculated from a minimum value of 1.0% RH is temperature dependent but so far this translates into a minimum of 0.17 mg.L
From a compressor enclosure air starting AH of 10.75 mg/L, a 0.17 mg/L is a 98.4% reduction in absolute humidity.

I went back and played around mathematically with the rate of change of the AH after I switch the air feed to the sensor from pre to post filter.
As I said the AH values drop quite rapidly and and starts to change in an inverse exponential pattern before all of a suddenflat-lining after about 90s.
Using a bit of maths and small amount of extrapolation it's possible to see how the AH will be even lower that what is disputed, maybe even as low as 0.05 mg/L.
If that is the case then its effectively removing at east 99.5% of the water.

When I GRTI I will do some more large volume air release tests, the extra significant digit in the AH should at least be useful to more clearly detect any changes in AH while air is being released.

[Michael G]

It may be more useful rather than sensing the pre-filter pressure (which after all is going to be pretty much atmospheric??) to get a differential pressure across the filter. At least that way you will have a ready reckoner on when the filter needs to be changed/ cleaned.

Michael

[BobL]

It may be more useful rather than sensing the pre-filter pressure (which after all is going to be pretty much atmospheric??) to get a differential pressure across the filter. At least that way you will have a ready reckoner on when the filter needs to be changed/ cleaned.

Michael

Sorry that's what I meant I'll be measuring the pressure difference across the filter but the cheap ($8) pressor sensors I've ordered measure relative to atmosphere so I'm going to be using two of them and comparing them.

The prefilter pressure is the same as the air pressure at the compressor 135 v 133 psi after recharge. The water sediment filter and Silica gel filter look like the lose another 5-6 psi.

[BobL]

I purchased a couple of cheap (~$13 each) 10 bar pressure sensors from ebay and decided I would try them on my compressed air filter/dryer.
378183

https://www.ebay.com.au/itm/1-8NPT-Stainless-Steel-Pressure-Transducer-Sender-Sensor-DC5V-Oil-Fuel-Air-Water/183493490807?_trkparms=aid%3D111001%26algo%3DREC.SEED%26ao%3D1%26asc%3D20160908105057%26meid%3D27acd80714764cb499a3d19f1a25dd37%26pid%3D100675%26rk%3D1%26rkt%3D15%26sd%3D183493490807%26itm%3D183493490807&_trksid=p2481888.c100675.m4236&_trkparms=pageci%3A9205b4f6-32f3-11e9-8aa6-74dbd180ae35%7Cparentrq%3Afd2bc9ee1680aa48895a21f4ffea0fbf%7Ciid%3A1

At that price I was not expecting too much but they each came with a reasonably rugged water resistant plug and socket that would cost that much alone here in Oz.

The first thing I did was to calibrate them to suit my Arduino micro controller so I set up the following Xmas tree on my portable compressed air supply (An old BBQ gas bottle).
BV is a bleed valve, Sensors are S1 and S2 and M is a 150 psi rated gauge.

The BV is connected to a Nitto style plug so can easily be removed and replaced with a oneway valve with Nitto style plugs on each end. Connecting up my compressor to the oneway valve and opening the valve on the gas bottle allows compressed air to enter the bottle. Then the bottle valve is closed, compressor disconnected and the one way valve is replaced by BV.

378184

With the micro controller running and displaying the Analog/digital Converted outputs (ADC's, numbers between 0 and 1023) of the sensors I slowly open the bleed valve so it loses ~5psi a minute and as the pressure drops simultaneously record the gauge pressure and the ADC's for each sensor ever minute or so. Then plot up a graph of Gauge pressure versus ADC for each sensor. The allows me to generate a calibration curve which is then programmed into teh micro controller so that it can display pressure (sorry for switching to kPa)

378185

I was really surprise at how linear their response is and how close together (<0.5%) their response slopes are, especially given I could really only read the gauge to +/- 1 psi and accuracy wise who knows where this really is.

Then I fitted a sensor either side of the filter/desiccator but the one on the downside stopped working after a few minutes. Given the burst rating is at least 2X the max pressure I doubt it was the problem. I did drop that sensor just before I installed it but I am thinking that maybe the moisture in the pre-filter side of the air might have had something to do with it n longer operating. The ADC output it gives at all pressures is "zero" which is similar to a shorted signal to ground response. This is odd because the sensor to ground resistance is ~15kΩ which is still the same as the working sensor.

I have tried a few things like sucking on the sensor with a vacuum pump and leaving it in the sun to see if it dries out but no change. Whether this is is just the vagaries of purchasing a cheap sensor or mistreatment or . . . .

Anyway not to worry too much as I already have small gauges on either side of the filter and the real reason for a pressure sensor is to see if I can calibrate the Humidity sensor under pressure with the aim of measuring the compressed air humidity while retaining it within the compressed air system compared to the current setup where the compressed air used to measure the humidity is lost from the system.

[BobL]

Just to round things out for the foreseeable future the low volume (0.1 to 3.0 L/Min) flow rate regulator came today and was installed as shown in the photo below.
It's the green coloured acrylic cylindrical object with the black adjustment knob on it similar to an Ar regulator.
The flow is indicated by a floating black bead hovering around the 1 L/min mark
378207
Accurate absolute humidity measurements have to be performed at low (<1 L/min) flow rates which turns out to be a faint whisper of air leaving the regulator exit.

Another photo of the whole shebang.
378208

For the first time I am starting to see the relative humidity of the air coming out of the filter going above the minimum reading of 1%. Today its 3% suggesting the desiccant is starting to load up, It will be interesting to see how long it takes to get to intake levels. I have been doing a lot of testing which consumes a lot of compressed air and probably does not help the readings..

[Poloris]

The low tech fix for reducing water content in compressed air lines is to put at least 10 meters of copper between the compressor and the water trap.
The arduino looks far more sexy than plumbing.
Mark

[BobL]

The low tech fix for reducing water content in compressed air lines is to put at least 10 meters of copper between the compressor and the water trap.
The arduino looks far more sexy than plumbing.
Mark

I actually have a finned copper heat exchanger similar to the one you show in the photo.
378352

Two Issues -
1) no spare area of wall large enough to hang it up on inside shed although I could hang it on an outside shed wall
2) No taps - would have to work out a way of removing the fins and leaving a smooth enough surface for tap junction. If I could work out way to get a smooth surface I could use two elbows, a Tee and a tap at the bottom of every U bend.

[QC Inspector]

Bob if you rotate that assembly 90º clockwise and spread the tubes enough so they zig zag down with a bit of slope you'll only need a single trap and valve at the end.

It could also be mounted under a ceiling or shelf as long as the outlet was sloped down from the inlet several degrees so the water would run to the trap.

No cutting and fewer fittings for either.

Pete

[BobL]

Bob if you rotate that assembly 90º clockwise and spread the tubes enough so they zig zag down with a bit of slope you'll only need a single trap and valve at the end.

Good idea.


It could also be mounted under a ceiling or shelf as long as the outlet was sloped down from the inlet several degrees so the water would run to the trap.
As it so happens I had a look outside the shed and there is an area of outside wall which could be used,
X marks the spot and the arrow shows the compressed air line coming out of the compressor enclosure.

378373

[QC Inspector]

Happy to give something back.

Pete

[BobL]

I had another look at the heat exchanger and an a bit wary about instigating cracks in the bends when stretching out the shape. I will ask my mech engineer BIL who gave it to me about its pressure rating. It comes from a solar HWS he and a colleague invented. I might try annealing the bends and see how easy it is to stretch out, if it's still remains under tension to stretch it out I probably wont use it.

[BobL]

Following on from QCI's idea . . . . . .

I finally managed to contact the Solar Engineer BIL who gave me this heat exchanger a few years ago and he said when new this finned copper tubing was rated at 25 bar so I figured running at 8 bar might be OK.
378559

I plugged one end up and connected it to the compressed at 8 bar and turned the air pressure on remotely - no bang, and pressure stayed good for 5 days.

Then I had to see if the 180º rounded corners would bend back apart to create angled horizontal runs - no problem (see below) I could bend them by hand more than enough.

The I needed to bend the short section marked with the red arrow through 90º so it was no longer bent.
There was no way this could be bent easily so I heated the corner with a propane torch and then stuck a 300 mm long steel bar partially down inside tube and it bent easily enough.

Mounting on the outside wall of the shed was tricky because side on the heat exchanger was quite floppy so I made up some temporary brackets/spacers to hold everything at the right angles. I had found these beaut little pipe wall offsets in my stash that were made for 19mm pipe (the 1/2" pipe plus fins makes the ID of the heat exchanger pipe 23mm) but there was enough meat in the green plastic material to allow for a 23 mm opening to be drilled.

Red arrow in this one shows where compressed air line enters shed.
378560

View the other way. You can just see the compressor enclosure and the red arrow in this one points to a temporary manual drain.
I will replace that with some sort of a solenoid I can activate inside the shed although it might be easier to allow any water to drain back into the tank as this already has an automated solenoid on it. Think I might need something different though.
378561

Haven't had time to test it though - did some chainsaw work for youngest bro this morning and I'm stuffed.

[BobL]

Last year I recharged (heated in an oven for 30 minutes) the desiccant in the comp air Humidity trap about every two months and it kept working although I was noticing a slow but steady degradation in the comp air humidity levels. When I broke my ankle late last year (6 weeks to get diagnosis) I was then stuck in a moon boot and I had to keep elevated for 7 weeks, shed work stopped - I turned every thing off and it took me a while to get back into the shed.

When I got back into the shed and turned everything back on I did not notice the compressor auto vent system had stopped working. I kept using the compressor and eventually noticed the humidity of the compressed air was effectively the same as the outside air. First thing was to repair the auto vent system - the Real time clock module has vibrated its way out of its socket- and a cup of water came pouring out of the vent before the desiccant (must put a manual tap on that as well) - opening up the desiccant trap I found it the moisture abasing gel was bight green (should be brown orange - see below) and almost dripping with water.


https://metalworkforums.com/attachment.php?attachmentid=377910&d=1548905853

So back into the oven for a recharge - all good.
However, when I put it back the comp air humidity was nowhere near as good as it had been last year. Back into the oven for a longer - hotter bake - still not much better!!! I did it a couple of more times - still no good

The other day I realised that the white poly sediment filter inside the humidity trap that prevents the SiGel beads from being blown down the comp air line (see orange arrow above) had never been dehumidified. Being poly it can't be cooked really hard in an over but I baked for a couple of hours at 75ºC and it certainly did make a big difference - still not quite as low an air humidity as when the poly filter was new but much better than before.

I'll try a longer time and maybe 5-10º warmer bake to see what that does.