[ftir_tga] FW: FTIR intake numbers & memory effect

  • From: David Griffith <griffith@xxxxxxxxxx>
  • To: "ftir_tga@xxxxxxxxxxxxx" <ftir_tga@xxxxxxxxxxxxx>
  • Date: Mon, 12 Dec 2011 15:28:56 +1100


From: David Griffith
Sent: Monday, 12 December 2011 14:38
To: 'Johannes Laubach'
Subject: RE: FTIR intake numbers & memory effect

Dear Johannes
I think your analysis is pretty good - while I am a bit surprised that Pevac 
comes out as high as 4 mb, and the error at 2 mb, we have never actually done 
anything to check it - we don't have any pressure gauge in that low pressure 
region.  But we DO know that the zero of the gauge is temperature dependent.  I 
will pass this on to Ecotech for the future - they should be able to measure 
this low pressure with more accuracy.  The other option would be to use a high 
vacuum pump to evacuate the cell to < 0.1 mb with confidence, and then zero the 
pressure gauge.

I'm not convinced that the 900-898 mb difference is due to the gauge though -  
after a fill, the pressure can readjust after the valves close for a static 
measurement.  The PID that controls the pressure uses the indicated pressure as 
it's control, so if the gauge reads 900mb, that is what it will adjust to, 
whatever the true value may be.

Cheers
David


From: Johannes Laubach [mailto:LaubachJ@xxxxxxxxxxxxxxxxxxxxxx]
Sent: Friday, 9 December 2011 11:14
To: David Griffith
Cc: Graham Kettlewell; John Hunt
Subject: FTIR intake numbers & memory effect

Hi Dave,
thank you, it is switched off now.
Yes the Aux Inlet numbering was correct in our test data.

I've been thinking more about the "residual-fill" error due to inaccurate 
pressure reading at low pressure. I think the best way to figure it out would 
be to use two different, reasonably constant, air sources (2 cylinders, or 1 
cylinder and the air in a closed lab) and define a measurement sequence of 3 
successive fills of the cell with Air 1, then 3 fills with Air 2, then Air 1 
again etc. The first of the three fills should be subject to measurable memory 
effect, the 2nd and 3rd fill should not be (and thus give  indistinguishable 
results). A few repetitions of this should suffice to work out if the effect is 
consistent between gas species or not.

Not having done that yet, the closest I've come to such a test is to use the 4 
to 14 Nov data for N2O, which had very small gradients between the two 
ambient-air intakes (<1 ppb) and a relatively large difference, of 13 %, 
between tank and ambient. So every time I measured Intake 1 after tank air, I 
had (on average) 0.21 ppb higher N2O than when I measured Intake 1 after Intake 
2. This works out as follows:
X_prev = 370.5 ppb (tank)
X_true = 327.7 ppb (ambient)
X_meas-X_true = 0.21 (+/-0.02) ppb, where the given uncertainty combines the 
standard error of the mean of 317 measurements of [N2O] in Intake 1 after tank 
air with that of 1066 measurements of [N2O] in Intake 1 after Intake 2.
Hence:
f=(X_meas-X_true)/ (X_prev-X_true) = 0.0049 (+/-0.0005)
in excellent agreement with the estimates from linear regressions in my earlier 
notes, which range from 0.003 to 0.007 for the different species.
With p_cell=898 hPa, this gives p_evac = f p_cell = 4.4 hPa (compared to 2.36 
hPa usually recorded). I presume this is an offset, not a scale factor error. 
An offset of 2 hPa is even consistent with regularly finding p_cell recorded as 
898 hPa when the target pressure was actually 900 hPa(!).

I tried the same procedure for CO, where we know X_prev = 0, but ambient [CO] 
was too variable, giving
(X_meas-X_true) = -0.1 (+/-1.3) ppb which has too large relative error to be 
useful. And for the other gases, the concentration difference between tank and 
ambient was too small as well as ambient too variable.

Cheers
Johannes.

********************************************************
Johannes Laubach
Ph.  +64 - 3 - 321 9865 (direct dial)
email: LaubachJ@xxxxxxxxxxxxxxxxxxxxxx
Fax  +64 - 3 - 321 9998

Postal Address:
Landcare Research, P.O.Box 40, Lincoln 7640, New Zealand

Street Delivery Address:
Landcare Research, Gerald St., Lincoln 7608, New Zealand
********************************************************

From: David Griffith [mailto:griffith@xxxxxxxxxx]
Sent: Thursday, 8 December 2011 4:01 p.m.
To: Johannes Laubach
Cc: Graham Kettlewell
Subject: RE: FTIR intake numbers

Johannes,
Thnaks for the help - Graham has found and fixed the bug with the numbering.
Can you confirm that it was only in the main inlet numbering, not the aux inlet?
In any case you can turn it off, we replicated the problem on our test bench.
Cheers
David


From: Johannes Laubach [mailto:LaubachJ@xxxxxxxxxxxxxxxxxxxxxx]
Sent: Thursday, 8 December 2011 13:03
To: David Griffith
Cc: Graham Kettlewell
Subject: RE: FTIR intake numbers

Hi Dave, Graham,
do you still need to access our FTIR for testing? Please let me know when I can 
switch off.
Cheers
Johannes.

********************************************************
Johannes Laubach
Ph.  +64 - 3 - 321 9865 (direct dial)
email: LaubachJ@xxxxxxxxxxxxxxxxxxxxxx
Fax  +64 - 3 - 321 9998

Postal Address:
Landcare Research, P.O.Box 40, Lincoln 7640, New Zealand

Street Delivery Address:
Landcare Research, Gerald St., Lincoln 7608, New Zealand
********************************************************

From: David Griffith [mailto:griffith@xxxxxxxxxx]
Sent: Tuesday, 6 December 2011 12:40 p.m.
To: Johannes Laubach
Cc: John Hunt; Samuel Hammer; Benoit Wastine (benoit.wastine@xxxxxxxxxxxx); Dan 
Smale; Vanessa Sherlock; Graham Kettlewell
Subject: RE: FTIR temperature

Hi Johannes,

Meanwhile Graham has replicated the error with the mis-named channel numbers, 
and is working on it.


Dave


________________________________
Please consider the environment before printing this email
Warning: This electronic message together with any attachments is confidential. 
If you receive it in error: (i) you must not read, use, disclose, copy or 
retain it; (ii) please contact the sender immediately by reply email and then 
delete the emails.
The views expressed in this email may not be those of Landcare Research New 
Zealand Limited. http://www.landcareresearch.co.nz

Other related posts: