A dissertation submitted to the Faculty of Science, University of the
Witwatersrand, in fulfillment of the requirements for the degree Master of Science.
Johannesburg, 2017. / Multicomponent gas mixtures are inherently challenging to produce in the
laboratory because of matrix effects, boiling points and reactivity amongst other
factors. Therefore, methods must be continuously developed to control these
challenges. The purpose of this work was to study these complex gas mixtures to
improve their measurements with emphasis on the reduction of uncertainty. There
are three critical steps to be followed in gas metrology for primary reference gas
mixtures of the highest metrological level; purity analysis of source gases,
gravimetric preparation and verification/validation which includes stability testing.
Purity analysis of select source gases was quantified using various techniques. This
methodology incorporated the use of molar masses and their uncertainties in order
to obtain purity values for the chemical composition of gas mixtures. While many
preparation methods such as permeation and dynamic methods are available, a
static gravimetric method was used to prepare the complex stack and automotive
gas mixtures following International Standard Organisation: 6142-1. For the mole
fraction range of interest, four components (carbon dioxide, carbon monoxide,
sulphur dioxide and nitric oxide) excluding propane, were obtained from analysis
by non-dispersive spectroscopy techniques calibrated by several standard gas
mixtures of different mole fractions. Propane was analysed by a gas chromatograph
coupled with flame ionisation detection. Multipoint calibration was used to
evaluate the linearity or nonlinearity of the detector.
The final results for the stack gas mixture components showed an achievement of
0.4% to 0.8% percentage relative expanded uncertainty and 0.4% to 1.3% for
carbon dioxide depending on the matrix of the standard gas mixtures used, 0.5%
to 1% for propane, 0.8% to 1.8% for nitric oxide, 2% to 6% for carbon monoxide
and 0.3% to 2.3% for sulphur dioxide. One of the most important suppositions
drawn was the incidence of synergistic effects associated with calibration by nonrepresentative
standard gas mixtures when these were used for analysis for some
of the components of stack mixtures. To evaluate improvements in measurement
capability, the results of the current work were compared to the data of the
laboratory in 2008-2011 and there was an improvement in the measurement of
carbon dioxide, carbon monoxide, propane and nitric oxide. These improvements
are attributed to rigorous purity analysis of starting materials, reduction of
uncertainty and developments in measurement expertise. In this work, different
measurement and calibration methods were used to analyse the components of the
new stack gas mixtures. The stability of these components was evaluated by
analysing them at different times and the statistical D-test was used to check for
significant instability.
An unknown stack sample was compared with the standard gas mixtures prepared
for this work. In combination with same matrix and same concentrations, single
point calibration was found suitable for stack gas measurement. To reiterate the
concept of matrix effect, the results of carbon dioxide in a mixture containing
carbon monoxide and oxygen as well in nitrogen, were used to show how
differences in matrix often give erroneous results and same conclusions cannot be
made for different mixtures. While the data of this measurement was
unsatisfactory, an improved method developed for this type of emission
multicomponent was very successful.
Emission industries also require automotive primary reference gas mixtures. These
are equally important and complex multicomponent mixtures measured and
improved in this work. A very precise and repeatable single point method was
developed for the analysis of the components of automotive mixtures. The
repeatability of the gas chromatography method was 0.2% for oxygen, 0.1% for
carbon monoxide, 0.5% for carbon dioxide and 0.3% for propane. The percentage
relative expanded uncertainty was 0.4% for oxygen, 0.8% for carbon monoxide,
0.8% for carbon dioxide and 0.5% for propane. However, its limitation was the use
of different calibration gases for each analysis. This led to inconsistencies in the
calculated mole fractions, non-predictability and instability. A proficiency testing
scheme was coordinated by the laboratory for automotive emission as part of this
study. Given the complexity of the samples, the work aimed to check any
improvements that could be made to the capability of measurement over the years.
This new method using gas chromatography coupled with different detectors
(residual gas analyser) was successful in verifying the gravimetric values very
V
accurately. Finally, the results of the stack gas mixtures were ≤1% relative except
carbon monoxide and ≤1% for automotive mixtures. This work aimed to support
the emission industry by providing it with representative and accurate reference
gas mixtures, extend the accreditation scope of the laboratory and improve its
calibration and measurement capability for multicomponent gas mixtures. / LG2018
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/24984 |
| Date | January 2017 |
| Creators | Marebane, Prelly Mohweledi |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | Online resource (xxii, [207] leaves), application/pdf |
Page generated in 0.0026 seconds