Beruflich Dokumente
Kultur Dokumente
I. INTRODUCTION
(2)
TABLE I
CONSTANTS FOR THE VOIGT FUNCTION APPROXIMATION [9]
(5)
where
Table I,
is
(8)
is a the linestrength at a reference temperature of
where
296 K, obtained from the HITRAN 2008 database [10],
and
are the partition functions at the gas cell temperature
and the reference temperature respectively, is Planks constant
and
is the transitional lower state energy. The linestrength
has to be converted from the HITRAN value using the equation
(9)
where
is Avagadros constant and is the gas constant in
units of cm atmK mol .
A number of parameters are given in the HITRAN 2008 dataand . The database also
base [10], including
provides values for self- and air-broadening parameters and the
temperature dependence of the self-broadening parameter, however, no gas-broadening parameters are included for other gases.
For the methane experiments a bulk-broadening parameter has
been calculated using absorption signals of known gas concentration, described further in Section IV. As there was no comparative measurement process for the water measurements it was
not possible to calculate a bulk-broadening parameter, therefore
the air-broadening parameter from the HITRAN 2008 database
was used.
The theory described in this section provides all the necessary information for deriving theoretical absorption profiles.
A MATLAB program has been written that compares experimental signals to the derived absorption profiles and carries out
a least-squares fitting routine to obtain concentration and pressure measurements, described in detail in Section III.
III. TDLS GAS COMPOSITION MEASUREMENT SYSTEM
(6)
and
1355
is
(7)
1356
Fig. 2. In-line gas sensor, constructed from 316 Stainless Steel Swagelok
fittings.
If only one gas species is being analysed the single-mode pigtail from the appropriate laser is connected to a 50:50 optical
splitter. If both gas species are being measured then single-mode
pigtails from each laser are connected to an optical switch, providing selection between the laser outputs via two 5 V sinusoidal signals from the DDS board. The selected laser output
then passes through the 50:50 optical splitter. One arm of the
splitter transmits 50% of the light through an in-line gas cell,
and the second arm is used to send 50% of the light to a fibre
ring resonator for wavelength referencing [10].
The in-line gas cell, shown in Fig. 2, is constructed from 316
grade stainless steel to reduce corrosion caused by high water
vapour concentrations. Both the gas cell and the signal processing have been described in detail previously for methane gas
composition measurements in a dry atmosphere [5]. Therefore,
only modifications applied to the system for the measurement
of water vapour are detailed here.
For water vapour measurements an 1/8 gas sample line is
taken directly from the SOFC test rig. The concentration of
%; the line preswater vapour in the sample line was always
sure was 6.4 bar ; the corresponding dew point was 72 C. The
sample line was trace heated to approximately 100 C and kept
1357
Fig. 3. Comparison of the final theoretical fit for the bulk-broadening parameter, obtained using a least squares fitting procedure of an experimental absorption profile from the anode ejector with a known methane concentration
of 11.8%.
(10)
is the self-broadening parameter, taken from the
where
HITRAN2008 database,
are the molar fractions of
methane and the rest of the gas mixture respectively and m and
n are the methane and bulk temperature exponents respectively.
As this data has been taken near room temperature, the ratio
of T to T is 1, therefore the values of m and n provide a
negligible contribution to the overall broadening.
The bulk broadening parameter was calculated using a
sample from the anode ejector. The fitting procedure was
similar to that described in Section III, however, in this method
all variables were fixed except for
.
The bulk-broadening parameter was calculated on two occasions, separated by two months, and an example calculation is
shown in Fig. 3. On both occasions a sample from the SOFC
system was analysed, where the methane concentrations were
calculated using GC as 11.8% and 14.9%, giving bulk-broadening parameters of 0.1211 cm
and 0.1252 cm
respectively. The overall variation between the two results is approximately 3%, and this error is assumed to be caused by the error
in the pressure reading from the analogue pressure gauge.
If both bulk-broadening parameters are used independently
to calculate the methane concentration from a different location in the SOFC system we obtain final concentration values
of 0.586% and 0.603% respectively. The overall error between
these two measurements is 2% and the absolute concentration
error is 0.017%. The concentration of methane determined by
the GC at the same location was 0.58% with an absolute error
of
%. The concentration values obtained from both bulkbroadening parameters are within the bounds of the GC measurement accuracy. Finally, assuming a signal to noise ratio of
3, the calculated lowest measurable absorbance is
1358
Fig. 5. Methane concentration purge time through the in-line gas analyser as
methane is introduced into the SOFC test rig.
Fig. 6. The temperature of the gas cell as a function of time, recorded using a
TC-08 temperature logger, during warm up and isothermal operation.
In order to calculate the total purge time through the gas cell
a step response was fitted to the data in Fig. 5, of the form
(11)
where is the measured concentration,
is the final concentration, is the time in seconds and is the time-constant, calculated to be 50 s.
In order to ensure a true gas composition measurement for an
individual sample line it is necessary to purge a gas cell for 250
s, assuming 1% measurement accuracy is at a time of .
Fig. 4 provides the total purge time of the sample line and the
gas cell (540 s), whereas Fig. 5 only provides the purge time
of the gas cell (250 s). Therefore, the sample line purge time
can be calculated to be the difference in those purge times, i.e.,
290 s. The purging of the gas cell takes a significant amount
of time due to the much larger volume per unit length of the
cell, as compared with the sample line. This information is used
to ensure that following each change in sampling position or
incremental change in system operating conditions, the corresponding sample line measurement can be made with an appropriate time delay to ensure an accurate gas composition is
obtained.
V. WATER MEASUREMENTS
In addition to monitoring methane concentration from various points around the SOFC system, it is also important to measure the water concentration. This cannot be carried out directly
using GC as it is unable to directly measure water vapour concentration. Using the TDLS system it is possible to measure
the water vapour concentration in-situ. In order to ensure no
water vapour condenses upstream of the TDLS measurement,
the sample line from the SOFC cathode is trace heated and the
TDLS sensor is located as close to the SOFC as possible as
described in Section III. The temperature inside the cell was
recorded as a function of time during warm up and stable operation, and the results are shown in Fig. 6.
During isothermal operation the temperature varied by approximately 3 C between 65.568.5 C, providing an average
temperature value of 67 C. To estimate the contribution of the
temperature variation on the water concentration measurements,
concentrations have been calculated for a single absorption profile using three different temperatures, the maximum, minimum
Fig. 7. Recorded water spectra at 67 C, taken when the fuel cell was operating
at half-load, and a comparison with a theoretical profile.
and average temperature. The dry air line was chosen for this test
and the three calculated concentration values were 0.3745%,
0.3764% and 0.3759% respectively. The temperature difference
between the maximum and minimum temperature values accounts for a concentration variation of only 0.0019%. Using a
number of datasets for the dry air line the average water vapour
concentration was measured to be 0.367% with a standard deviation of 0.0141%. Therefore, the contribution of the temperature
variation is not the most significant factor in the concentration
measurement error. Finally, assuming a signal to noise ratio of
3 the lowest measurable absorbance has been calculated to be
using these datasets. This relates to a minimum
measurable water vapour concentration of 31 parts per million.
Water vapour concentration measurements were also taken
when the SOFC was operating at full-electrical load and halfload. At half-load the average water vapour concentration was
measured as 0.6378%, shown in Fig. 7, and at full-load the concentration was 0.83%; these values are proportional to fuel flow
rate at these specific test conditions.
The SOFC system must be able to operate under a range of
conditions similar to those that may be experienced by a commercial product. One of these is the moisture content of the
cathode. Fig. 8 presents the measurement of water vapour concentration as water is introduced into the SOFC cathode stream.
Measurements were taken at a rate of one measurement every
1020 seconds and four separate measurement runs were car-
1359
Michael Lengden received the M.Phys. degree in physics and the Ph.D. degree from the University of Manchester, Manchester, U.K., in 2001 and 2006,
respectively, where he was engaged in stepwise excitation of atomic and molecular metastable states.
He was an Applications Engineer at the Laboratory Impex Systems, Ltd.,
prior to becoming a Research Fellow in the Centre for Microsystems and Photonics at the University of Strathclyde between 2007 and 2010. He is currently
a Lecturer in the EEE Department, University of Strathclyde, Glasgow, U.K.,
where he is engaged in high-temperature spectroscopy measurements using tunable diode laser spectroscopy. His current research interests include new techniques in tunable diode laser spectroscopy for concentration and pressure measurements in harsh environments, such as solid oxide fuel cells and aeroengines.
Robert Cunningham received the B.Sc. (hons.) degree in chemistry from the
Universiy of Glasgow, Glasgow, U.K., in 1989, and the Ph.D. degree in catalysis
from Brunel University, U.K., in 1992.
From 1992 to 1995 he worked at Eindhoven University of Technology, The
Netherlands, developing a new chemical process for the production of positronemitter labelled alkanes at the universitys cyclotron facility. From 1995 to 2000
he was employed as a post doctoral research fellow at the University of Keele,
where he investigated alkane steam reforming in solid-oxide fuel cells. Since
2000 he has been employed by Rolls-Royce Fuel Cell Systems Ltd. where he
has responsibility for SOFC performance analysis.
Walter Johnstone received the B.Sc. (Hons.) degree in chemical and material
sciences and the Ph.D. degree in laser physics from the University of Strathclyde, Glasgow, U.K., in 1977 and 1982, respectively.
From 1980 to 1986, he held Project Engineering and Senior Technical Management posts at Pilkington Optronics, Ltd. (now Thales) and Logitech, Ltd. In
1986, he joined the University of Strathclyde where he is currently Vice Dean of
the Faculty of Engineering and a Professor of Photonic Systems in the Department of Electronic and Electrical Engineering. Since 1994, he has been a Director of OptoSci, Ltd., Glasgow, where he has been instrumental in bringing a
number of photonics products to market. His work has led to more than 200 technical publications and several patent applications. His current research interests
include optical waveguide components, fiber lasers, distributed fiber sensing
systems, and optical gas sensing.