Testing WVO and biodiesel for water.

[Julian]

Having thought about the above a little, if these sensors are that sensitive to methanol, then that could be quite a serious source of inaccuracy.

[Bill]

My bio was dried by what I call splash drying at 65 C. I then made up the 'wet' samples as ppm by weight of 200, 500, 800 ppm of water, mix thoroughly. After adding the water and giving a good shake to mix they all remained clear and no sign of free water!

Here are some basic stats from my measurements. I won't include any graphs because I haven't figured out how yet. These results show the overlap between sample readings which mean it would not be useful as a measurement tool in this form for the level of accuracy required. All measurements are output volts from the sensor.

Variable    N       Mean     Median     StDev     Minimum    Maximum
std(dried)  8     3.1175     3.1200     0.0271    3.0800     3.1600     
200ppm    8     3.1550     3.1700     0.0382    3.0700     3.1900     
500ppm    8     3.2337     3.2350     0.0311    3.1900     3.2800     
800ppm    8     3.2737     3.2800     0.0378    3.2300     3.3200     


When I get back of holiday I will try some samples in the carbide manometer for comparison.

[Julian]

Interesting.  Where you have a spread of readings, are they based on multiple tests or were the readings fluctuating on a single test?

I'll try and prepare samples similar to yours today and post the results.

[Bill]

Results are from 8 readings on each sample. Voltage measurement had stabilised at the second decimal place after 2 minutes but the third place was still up and down a bit. Values listed are to three places as a result of calculations.

[Julian]

Been playing a little more this afternoon.

First obstacle is that I don't seem to have the right equipment to make the samples.  My scales only go to 0.1g and I don't really have the facilities to get a litre plus of bio up to high temperatures ... wife is very protective of her kitchen equipment!  The scales wouldn't even weigh out 0.1g as they appeared to require something exceeding that weight to set them operating.

Anyhow, she was out today so I did manage to get access to the microwave.  I heated some bio to 130°C and set it running on the stirrer for a few mins., so I'm fairly sure it was a dry sample.  I then took 100ml of the bio and added 4 drops of water.  No idea what ppm that gave, but it can't be too much.

I then ran the following measurements ...

"Dry"
Stablised out at 26.6% @ 44.3°C

"Wet"
Stablised out at 49.4% @ 41.6°C

Added another 4 drops of tap water to the wet sample ...
Stablised out at 68.3% @ 41.7°C

Back to "Dry" sample
This had cooled by now so I brought up the temperature on the hot plate to match the initial reading and got ...
27.2% @ 44.3°C

Still not at all scientific but it seems to indicate a reasonable degree of repeatability and stability.

I think I need to look at some better scales and/or a new wife.

Alternatively, if anyone with suitable lab equipment wants to replicate the Arduino set up I'd be happy to detail what to buy.  As I said previously, it should be less than a tenner.

[dgs]

Julian, I think your results are really good, very encouraging.

4 drops of water from a 1ml pipette is about 0.12mls in 100mls of bio is 1200ppm. Of course we need to be looking at concentrations of say 150ppm to 1200ppm in finished bio or in the process of being dried. From samples of bio that I tried to dry when I was calibrating the manometer I found it needs to be left with heat far longer that a few mins.

Nevertheless, that is a good set of experimental tests. it may help when I get my meter to send you down some samples that have been checked with the Sandy Brae or KF. Or you could send me some. I don't think in this case the KF standards of 100ppm water would be of any use as I think they are in ethanol.

In the meantime could you send me the Arduino details, many thanks, Dave.

[Julian]

Yes, I think they look promising, but need a whole lot more work.  I'm trying to get my head round the need for the temperature reading.  I think I can see why it's needed with an air measurement, but I'm struggling to see the need in bio or oil.

I'll have another play tomorrow armed with official British Standard water drop volume and perhaps dry for longer (she's out again tomorrow) and try one drop in 200ml.  If it distinguishes between that and a dry sample, it must be worth perusing.

Details of the Arduino set up I'm using are ...


Sensor

This is the sensor a DHT22 I've been playing with ... https://learn.adafruit.com/dht/overview
loads available on ebay circa 2-3 quid.

Connect to the Arduino board using a 10K resistor thus ...




Arduino

I've been using a UNO, there are other types, possibly cheaper, but the UNO is standard stuff.

Chinese copies available on ebay.  Typically this one ... http://www.ebay.co.uk/itm/High-Quality-1Pc-ATmega328P-UNO-R3-Board-with-USB-Cable-for-Arduino-/141513539379?pt=LH_DefaultDomain_3&hash=item20f2dd3f33   £4.00 with USB lead.


Arduino IDE and libraries

Free downloads,
IDE from ... http://www.arduino.cc/en/Main/Software
LCD library ... http://www.arduino.cc/en/Reference/LiquidCrystal?from=Tutorial.LCDLibrary
DHT library ... https://github.com/adafruit/DHT-sensor-library
The Arduino sketch to drive both the LCD and the sensor can be copied and pasted from earlier in this thread or I'll happily send you the one I'm currently playing with.

LCD

I used this type because I had it and because it plugs straight into the UNO board (also plugs into the Mega board too) ... http://www.ebay.co.uk/itm/1602-LCD-Board-Keypad-Shield-Blue-Backlight-For-Arduino-Expansion-Board-M1-/141676880566?pt=LH_DefaultDomain_3&hash=item20fc99a2b6  currently £1.65 with one day left.

A little more complicated to rig up but you could use this ... http://www.ebay.co.uk/itm/HD44780-Backlit-1602-LCD-Display-Blue-16x2-Arduino-Raspberry-Pi-AVR-/251540198402?pt=LH_DefaultDomain_3&hash=item3a90f6d402 £2.99 BIN, but you'll need wires, a resistor and a 10K pot, connected like this ...



Just noticed that this diagram uses pin 2 so the Arduino sketch will need altering to run the sensor off a different pin ... easy to do.

Battery pack
to make it portable typically ... http://www.ebay.co.uk/itm/AA-Battery-Batteries-Holder-Box-with-DC-2-1-Power-Jack-for-Arduino-Tracing-Car-/400701166131?pt=LH_DefaultDomain_3&hash=item5d4ba69233
or
http://www.ebay.co.uk/itm/PP3-9v-Battery-Clip-to-5-5-2-1mm-Barrel-Jack-for-Arduino-and-CCTV-/111619871728?pt=LH_DefaultDomain_3&hash=item19fd1013f0


Kits as an alternative
You can get Arduino kits typically ... http://www.ebay.co.uk/sch/i.html?_odkw=arduino+experiment&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR7.TRC2.A0.H0.Xarduino+kit.TRS0&_nkw=arduino+kit&_sacat=0

But check they have everything you need, many don't contain the humidity sensor.

[Julian]

For information, I've just found the manufacturers spec sheet for the DHT22 or AM2302, as it's sometimes called.

http://akizukidenshi.com/download/ds/aosong/AM2302.pdf

[Tony]

Having thought about the above a little, if these sensors are that sensitive to methanol, then that could be quite a serious source of inaccuracy.

Sounds like a serious source of checking your bio is meth free to me

[Julian]

Having thought about the above a little, if these sensors are that sensitive to methanol, then that could be quite a serious source of inaccuracy.

Sounds like a serious source of checking your bio is meth free to me

Glass half full?

But if the sensor can't distinguish between methanol and water, it's not a lot of use testing for either.  You would have to make the assumption that any water washing had removed the methanol.  I don't think you can assume dry oil and ASM will produce water free bio, can you?

I haven't tried the sensor in veg oil yet,  but obviously methanol won't be an issue testing that.

[dgs]

I woudn't be too worried about methanol as if testing bio we are either testing water washed or de-methed then well settled.
Either way there should not be any methanol present.
On a personal note my bio will not be tested until at least half way through it's drying cycle. This test is really advantageous because of it's convenience, if it gives results only accurate to +/- 50ppm it is more than accurate enough.

We are really finding out the range our water level is at. Ideally we should all be trying to get that level down to <300ppm (the German spec)
After using my turbo dryer I usually get a result of about 150ppm.

[Julian]

A bit of good news this morning.

I got a very prompt reply from IST AG in Switzerland, having emailed them last night.  IST AG are the manufacturers of the sensor I flagged up a while back as possibly being suitable for use in oil.

It transpires these sensors are designed to be immersed in oil and monitor the moisture content.

The very helpful chap who replied attached an up dated data sheet and said that application notes were being prepared but until they are ready he'll happily answer any questions we have.

It appears that the new data sheet is too big to attach to a post ... strange as it's only a two page pdf.  However if anyone wants a copy, email me and I'll forward it.

[Tony]

But if the sensor can't distinguish between methanol and water, it's not a lot of use testing for either.

I would argue that that doesn't matter, I don't want either in my bio and the mechanism for removing both is the same.

[Julian]

Fair point.

I wonder if anything else will effect it ... I can try acetone and IPA fumes.

[Julian]

Done a little more playing over the last couple of days ...

Ambient air 47.7% @ 17.7°C
IPA fumes in a beaker 31.75 @ 17.4°C
Back to ambient air 48.5% @ 16.4°C
Methanol fumes in a beaker 99.9% @ 15.4°C (reached this figure very quickly)
Back to ambient air 58.2% @ 16.6°C (took 30 mins to reach this reading, recovering very slowly)

Also ...

Dried a sample of bio for around 10 mins at 130°C.  Took a 200ml sample and added 0.05ml tap water, so not withstanding SG that's approximately 250ppm.  Started testing when bio had cooled to a reasonable temp with the following results ...

	Time	Duration in sample	%RH	Temp°C	Check Temp°C	Notes
Wet	00:00:00	?	35.6	21.4	22.0
Dry	00:00:00	?	26.4	22.3	22.6
Wet	14:49:00	?	40.1	18.0	18.6
Dry	14:58:00	00:09:00	29.5	19.2	19.5
Wet	15:17:00	00:19:00	41.9	17.2	17.7
Dry	15:36:00	00:19:00	32.1	17.7	18.2
Wet	16:13:00	00:37:00	43.5	16.4	17.0
Dry						Lost the figures … Muppet!
Wet	17:26:00	?	47.8	15.4	16.0
Dry	18:16:00	00:50:00	36.9	14.8	15.4
Wet	21:19:00	03:03:00	49.7	15.6	16.1
Dry	12:02:00	14:43:00	54.6	15.9	16.5	Next day
Wet	17:07:00	05:05:00	58.1	15.9	16.6
Dry	12:23:00	19:30:00	63.5	15.7	16.3	Next day
Wet	12:50:00	00:27:00	63.5	16.2	16.9
Ambient air	19:48:00	06:58:00	45.9	18.7	0.0

At the end of the tests when the %RH was coming out the same I left the probe in air to check it wasn't broken.  The readings after nearly 7 hours seem plausible, so I'm guessing both samples absorbed quite a bit of moisture from the air.

Bit of a mine field this moisture content business!