In this paper we developed an automatic sampling device which is capable of taking samples of the substrates at pre-programmed schedule and storing the samples for retrieval. For the reason that the sampler is operated in a severe ocean environment, the first design consideration is to use mechanisms and parts as simple as possible to reduce the risk of failure. Another primary concern is to make sure the design can preserve sediment securely in water. Besides, other considerations include size, weight, and functions of programmable sequential sampling. The sampling device is developed in two phases. Firstly, a device equipped with only one sampler is designed for achieving the purpose of programmable sampling. Secondly, the design will be extended to have multiple samplers to accomplish the task of sequential sampling. Six main designs are implemented in the first stage. They are the conceptual design of the grab for taking and storing sediment, the design of separation and reunion between grab and transmission mechanism, gearing design for transmitting motion of the sampler, timer and control circuit design for the lowest power consuming consideration, pressure hull design for installing a motor, batteries, and needed electric devices, and frame design for firmly supporting all instruments. The design of phase one was completed and its preliminary testing result came out successfully except that the collection chamber is not sealed securely enough. A new sample holder was designed and then successfully solved the problem of sediment sealing. According to the design of phase one, the concept design of the sequential sampling device of having 24 sample holders is presented.

The primary focus of this study was to compare the efficiency of six filter samplers in the collection of inhalable soluble platinum (Pt) salts at a South African base metal refinery. Inhalation remains the major route of occupational exposure to platinum groups metals (PGMs). South Africa would benefit from the study since it’s amongst the major countries where PGMs are produced and hence, monitoring of worker exposure with the most efficient sampler is of utmost importance. The IOM is currently being used in routine exposure monitoring although no studies have been carried out to compare its performance to that of the other samplers under the actual base metal refinery conditions. Method: The button, closed face cassette (CFC), Gesamtsstaubprobenhome (GSP), (Institute of Medicine) IOM, PAS-6 and seven hole (SH-sampler) samplers were randomly allocated to six different positions in presumably high exposure areas. The samplers were moved around in the subsequent sampling days and the process repeated 3 times. The average dust mass and Pt concentrations were used as a basis of sampler performance and comparisons from which sampler hierarchies were determined. Results: The average relative humidity ranged between 37% and 43% and the average dry bulb temperature of 22.4°C was measured. Comparison of the dust mass concentrations revealed no statistically significant differences amongst the six filter samplers tested. The SH-sampler and CFC however collected the highest and lowest dust mass and Pt concentrations respectively. Discussion: The SH-sampler was found to be a sampler with more reliability than the the IOM for the collection of dust mass and soluble Pt. The IOM collected 98% of the SH-sampler dust mass and Pt concentrations. This was in spite of the larger variations indicated by the highest relative standard deviations and confidence intervals shown by the IOM than the other samplers. The GSP sampler, however, showed better precision than all the other samplers in the collection of platinum. The seven 4 mm orifices of the SH-sampler sampler allow for uniform distribution of sampled particles onto the filter supporting its better precision than the IOM which has only one 4 mm opening. The worst performing sampler was the CFC sampler since it collected the lowest dust mass and Pt concentrations. The CFC and the PAS samplers have downward facing inlets that are affected by gravity especially in lower wind speeds which, therefore, influences their efficiency. The GSP sampler concentrations placed it as 4th and 3rd best in Pt and dust mass hierarchies respectively even though it showed better precision than SHS in the sampling of Pt. The button sampler did not perform as well as would have been expected considering that its many evenly spaced orifices and the stainless steel are meant to reduce sample losses. Conclusion: The sampler hierarchy according to dust mass concentrations was in the following order: SH-sampler, IOM, PAS, GSP, button and CFC. The hierarchy obtained from Pt concentrations gave the order as SH-sampler, IOM, GSP, button, PAS and CFC. Similar studies have to be undertaken in primary and secondary platinum workplaces to validate the study results. Such studies should compare better performing samplers (SHS, IOM, Button and GSP) as well as incorporate particle size determination and distribution in those areas. / MSc (Occupational Hygiene), North-West University, Potchefstroom Campus, 2014

The primary focus of this study was to compare the efficiency of six filter samplers in the collection of inhalable soluble platinum (Pt) salts at a South African base metal refinery. Inhalation remains the major route of occupational exposure to platinum groups metals (PGMs). South Africa would benefit from the study since it’s amongst the major countries where PGMs are produced and hence, monitoring of worker exposure with the most efficient sampler is of utmost importance. The IOM is currently being used in routine exposure monitoring although no studies have been carried out to compare its performance to that of the other samplers under the actual base metal refinery conditions. Method: The button, closed face cassette (CFC), Gesamtsstaubprobenhome (GSP), (Institute of Medicine) IOM, PAS-6 and seven hole (SH-sampler) samplers were randomly allocated to six different positions in presumably high exposure areas. The samplers were moved around in the subsequent sampling days and the process repeated 3 times. The average dust mass and Pt concentrations were used as a basis of sampler performance and comparisons from which sampler hierarchies were determined. Results: The average relative humidity ranged between 37% and 43% and the average dry bulb temperature of 22.4°C was measured. Comparison of the dust mass concentrations revealed no statistically significant differences amongst the six filter samplers tested. The SH-sampler and CFC however collected the highest and lowest dust mass and Pt concentrations respectively. Discussion: The SH-sampler was found to be a sampler with more reliability than the the IOM for the collection of dust mass and soluble Pt. The IOM collected 98% of the SH-sampler dust mass and Pt concentrations. This was in spite of the larger variations indicated by the highest relative standard deviations and confidence intervals shown by the IOM than the other samplers. The GSP sampler, however, showed better precision than all the other samplers in the collection of platinum. The seven 4 mm orifices of the SH-sampler sampler allow for uniform distribution of sampled particles onto the filter supporting its better precision than the IOM which has only one 4 mm opening. The worst performing sampler was the CFC sampler since it collected the lowest dust mass and Pt concentrations. The CFC and the PAS samplers have downward facing inlets that are affected by gravity especially in lower wind speeds which, therefore, influences their efficiency. The GSP sampler concentrations placed it as 4th and 3rd best in Pt and dust mass hierarchies respectively even though it showed better precision than SHS in the sampling of Pt. The button sampler did not perform as well as would have been expected considering that its many evenly spaced orifices and the stainless steel are meant to reduce sample losses. Conclusion: The sampler hierarchy according to dust mass concentrations was in the following order: SH-sampler, IOM, PAS, GSP, button and CFC. The hierarchy obtained from Pt concentrations gave the order as SH-sampler, IOM, GSP, button, PAS and CFC. Similar studies have to be undertaken in primary and secondary platinum workplaces to validate the study results. Such studies should compare better performing samplers (SHS, IOM, Button and GSP) as well as incorporate particle size determination and distribution in those areas. / MSc (Occupational Hygiene), North-West University, Potchefstroom Campus, 2014

The act of dust sampling to characterise potential hazards in the respiratory tract is a process that requires initial removal of non-inspirable particles followed by separation into regional fractions. Particles that are not inhaled cannot deposit anywhere in the respiratory system. Furthermore, there is an important category of aerosol for which the conditions for potential injury is inhalation and deposition anywhere in the respiratory tract. Thus inspirable dust sampling is fundamental to all aerosol sampling for health risk evaluation. A proper estimation of inhalation exposure will need to examine the relationship between the inspiration efficiency of a sampler and particle size distribution of a dust cloud; and methods to collect a representative sample in the workplace.

No description available.

Two experimental configurations of linear slot nozzle virtual impactors were constructed and experimentally investigated for use as bioaerosol concentrators. In one configuration, the Linear Slot Virtual Impactor (LSVI), the nozzle was a straight slot having a length of 89 mm (3.5"). In the second configuration, the Circumferential Slot Virtual Impactor (CSVI), the nozzle was curvilinear following a circular path having a diameter of 152.4 mm (6.0") and the resulting total slot length was 479 mm (18.8"). Multiple prototypes of the two configurations were constructed having nozzle widths that varied from 0.508 mm (0.015") to 0.203 mm (0.008"). Optical and physical measurements were made of the nozzle dimensions in the critical region of the virtual impactor units. For the LSVI units the misalignment between the acceleration nozzle and the receiver nozzle was measured between 6 µm (0.00025") and 29 µm (0.00114"). This represented a range of 2% to 10% misalignment relative to the acceleration nozzle width. The CSVI Unit 1 and 2 misalignments were measured to be 15 µm (0.00061") and 9 µm (0.00036"), or 10% and 1.8% relative misalignment, respectively. The virtual impactors were tested with liquid and solid monodisperse aerosol particles. For operation at flow rate conditions predicted from the literature to produce a cutpoint of 0.8 µm AD, an acoustic resonance was observed, corresponding to significant nozzle wall losses of particles and an absence of normal particle separation in the virtual impactor. The onset of the resonance phenomenon was observed to begin at a nozzle Reynolds number of approximately 500 for the LSVI configuration, and 300 for the CSVI configuration. For flow rates just below the onset of resonance, normal virtual impactor behavior was observed. The value of Stk50 was 0.58 for both devices, corresponding to a particle cutpoint size of 1.1 µm AD for the LSVI configuration and 2.2 µm AD for the CSVI. The collection efficiency was greater than 72% for all particle sizes larger than twice the cutpoint up to the largest particle size tested (≈ 10 µm AD). The peak collection efficiency for both concentrators was greater than 95%.

SARS-CoV-2 wastewater surveillance is a promising tool for monitoring the spread of infection during pandemic outbreaks. 24-hour composite sampling of wastewater using autosamplers is the preferred means for wastewater surveillance sample collection. Autosamplers however require a significant capital cost and furthermore some sampling locations are not amenable to autosampler deployment because of a lack of space and lack of access to electricity. Grab sampling is an alternative to auto sampling for wastewater surveillance, however it may be less effective compared to 24-hour composite sampling due to the possibility to miss the collection of shed disease targets during critical shedding events. Torpedo-style passive samplers packed with medical gauze and tampon-style passive samplers are alternatives to grab sampling when deployment of autosamplers is not possible. Torpedo-style and tampon-style passive samplers are characterized as being easy to deploy and collect and have shown promise for disease surveillance using wastewater. Although passive samplers have shown the ability to detect SARS-CoV-2, they have not demonstrated the ability to quantify the viral load in the wastewater due to the fact that the collection of the liquid phase of the sampler is not consistent across the deployment period of a passive sampler. As SARS-CoV-2 disease targets have been shown to largely partition to the solids phase of wastewaters, it is hypothesized that mass fraction quantitation may enable passive samplers to quantify wastewater signals comparably to autosamplers. In this study, wastewater samples were collected from the same location over a period of three months from a sewer access point at the University of Ottawa using conventional 24-hour auto sampling. Two types of torpedostyle passive samplers and a tampon-style passive sampler were tested to assess whether passive sampler measurements of SARS-CoV-2 N1 and N2 gene targets can be used in the place of autosampler quantitated values. When comparing the wastewater characteristics of centrifuged pellets collected by various passive samplers and a conventional autosampler, the results of this study showed that the torpedo-style passive sampler packed with two pieces of gauze (P2) collected significantly lower water content compared to the autosampler, and P2 collected significantly greater total solids and volatile solids compared to the autosampler. When measuring SARS-CoV-2 N1 and N2 signals, the results indicate that N1 and N2 gene region copy numbers from all of the samplers were not significantly distinct. However, the P2 sampler, a torpedo-style passive sampler packed with four pieces of gauze (P4), and the tampon-style passive sampler (T) captured a greater quantity of pepper mild iii mottle virus (PMMoV) gene targets compared to the autosampler; where PMMoV is the most commonly measured fecal biomarker for wastewater surveillance of SARS-CoV-2. The greater quantity of PMMoV gene targets compared to the autosampler was likely due to proportionally higher total solids and volatile solids in the centrifuged pellet material captured. When N1 and N2 measurements were normalized against sample volume, pellet mass or PMMoV gene copy numbers, P2, P4, and T showed no significant differences compared to the autosampler. In contrast, differences were observed between passive samplers and the autosampler when PMMoV measurements were normalized against the matrix volumes or pellet mass. High statistical percentage differences were observed between all passive samplers and the autosampler. Overall, passive samplers are reliable, cost-effective devices for sampling disease targets in wastewater if results are expressed as copies/g or copies/copies PMMoV. These devices are feasible substitutes for autosamplers when detection and quantification of SARS-CoV-2 in wastewater are required. P2 passive samplers using units of measurement of copies/g are recommended for SARS-CoV-2 surveillance in the wastewater.

Bedload-trapping efficiencies for four types of pressure-difference bedload samplers – a standard Helley-Smith (intake-nozzle width and height of 76.2 mm x 76.2 mm), BLH-84 (76.2 mm x 76.2 mm), Elwha (203 mm x 102 mm) and Toutle River-2 (305 mm x 152 mm) a standard Helley-Smith, US BLH-84 (both with intake nozzle dimensions of 76.2 mm × 76.2 mm), Elwha (203 mm × 102 mm) and Toutle River-2 (TR-2; 305 mm × 152 mm) – were calculated from data collected during the StreamLab06 experiments in the St. Anthony Falls Laboratory Main Flume during January-March 2006. Sampler nozzle-flare ratios –the area of the nozzle's outlet divided by its inlet area – equaled 1.4 for all but the Helley-Smith sampler's nozzle-flare ratio of 3.22. A sampler's trapping coefficient quantifies its bedload-trapping efficiency. Technically supportable trapping coefficients are divided into raw trapping rates measured by the sampler to produce "true" bedload-transport rates equivalent to that which was inferred to have occurred in the absence of the sampler. Six combinations of sampler and bed types were tested; the BLH-84, Elwha, and Helley-Smith samplers were deployed atop a sand bed (D50 = 1.0 mm) during five steady flows ranging from 2.0-3.6 m3/s. The BLH-84, Elwha, and TR-2 samplers were deployed atop a gravel bed (D50 = 11.2 mm) at four steady flows ranging from 4.0-5.5 m3/s. Thirty-seven trials – repeated manual at-a-point deployments of a given bedload sampler for a given steady flow and bed type – took place. Trapping coefficients were calculated for each sampler and bed type in which it was deployed. Ergo, two of the samplers – the BLH-84 and Elwha – were each assigned two trapping efficiencies for sampling on a sand versus a gravel bed. These data were evaluated using four analytical methods: Ratio of Averages: This relatively simple and straight-forward method required calculating averages of bedload-transport rates derived for each of the 37 trials for a given bedload sampler and for up to nine combinations of weigh pans and time intervals. The computations were performed using untransformed data. Average of Ratios: This more complex method using real-space trapping data involved developing average transport rates from selected pan data for each bedload sample. Pan transport-averages were calculated for each interval equal to the duration of a single at-a-point bedload measurement, ranging from 15-180 seconds. Ratios (coefficients) were calculated by dividing each interval average into the single-sample trap rate. Those ratios were then averaged to produce a single trapping coefficient for the trial and then combined into a single average for each bedload-sampler/bed type/flow combination. Modified Thomas and Lewis Model (1993): The Thomas-Lewis Model was revised to operate using untransformed data in addition to cube-root transformed data (thus, the third and fourth analytical methods used, respectively), and to use nine pan-window combinations to calculate trapping coefficients. The original 3-step model required first regressing cube root-transformed sampler data on time-window averaged pan transport rates. The second step squared the regression residuals from the first step on the variance of the cube root of the interval-mean transport rate for the time window. The predicted values from the second-step regression were inverted and used as weights to re-estimate the first-step regression. Generalized trapping-coefficient calculations based on results from the four analytical methods for the bed-types in which the samplers were deployed follow: • BLH-84 Sampler: A 0.83 sand-bed trapping coefficient and 0.87 gravel-bed coefficient, which could be averaged to a single coefficient of 0.85. • Elwha Sampler: A 1.67 sand-bed trapping coefficient and 1.54 gravel-bed coefficient, which could be averaged to a single coefficient of 1.6 • Helley-Smith Sampler: The 3.11 sand-bed trapping coefficient could be applied as such or reasonably simplified to a value of 3.0, and • TR-2: The gravel-bed trapping coefficient equaled 1.70. An unadjusted bedload-trapping rate calculated from a sample collected by a given sampler can be divided by its trapping coefficient(s) to obtain the most reliable transport-rate value. / Ph.D.

Bayesian subset selection suffers from three important difficulties: assigning priors over model space, assigning priors to all components of the regression coefficients vector given a specific model and Bayesian computational efficiency (Chen et al., 1999). These difficulties become more challenging in Bayesian quantile regression framework when one is interested in assigning priors that depend on different quantile levels. The objective of Bayesian quantile regression (BQR), which is a newly proposed tool, is to deal with unknown parameters and model uncertainty in quantile regression (QR). However, Bayesian subset selection in quantile regression models is usually a difficult issue due to the computational challenges and nonavailability of conjugate prior distributions that are dependent on the quantile level. These challenges are rarely addressed via either penalised likelihood function or stochastic search variable selection (SSVS). These methods typically use symmetric prior distributions for regression coefficients, such as the Gaussian and Laplace, which may be suitable for median regression. However, an extreme quantile regression should have different regression coefficients from the median regression, and thus the priors for quantile regression coefficients should depend on quantiles. This thesis focuses on three challenges: assigning standard quantile dependent prior distributions for the regression coefficients, assigning suitable quantile dependent priors over model space and achieving computational efficiency. The first of these challenges is studied in Chapter 2 in which a quantile dependent prior elicitation scheme is developed. In particular, an extension of the Zellners prior which allows for a conditional conjugate prior and quantile dependent prior on Bayesian quantile regression is proposed. The prior is generalised in Chapter 3 by introducing a ridge parameter to address important challenges that may arise in some applications, such as multicollinearity and overfitting problems. The proposed prior is also used in Chapter 4 for subset selection of the fixed and random coefficients in a linear mixedeffects QR model. In Chapter 5 we specify normal-exponential prior distributions for the regression coefficients which can provide adaptive shrinkage and represent an alternative model to the Bayesian Lasso quantile regression model. For the second challenge, we assign a quantile dependent prior over model space in Chapter 2. The prior is based on the percentage bend correlation which depends on the quantile level. This prior is novel and is used in Bayesian regression for the first time. For the third challenge of computational efficiency, Gibbs samplers are derived and setup to facilitate the computation of the proposed methods. In addition to the three major aforementioned challenges this thesis also addresses other important issues such as the regularisation in quantile regression and selecting both random and fixed effects in mixed quantile regression models.

Time-synchronization in wireless ad-hoc sensor networks is a crucial piece of infrastructure. Thus, it is a fundamental design problem to have a good clock syn- chronization amongst the nodes of wireless ad-hoc sensor networks. Motivated by this fact, in this thesis, the joint maximum likelihood (JML) estimator for relative clock phase offset and skew under the exponential noise model for the reference broadcast synchronization protocol is formulated and found via a direct algorithm. The Gibbs Sampler is also proposed for joint estimation of relative clock phase offset and skew, and shown to provide superior performance compared to the JML-estimator. Lower and upper bounds for the mean-square errors (MSE) of the JML-estimator and the Gibbs Sampler are introduced in terms of the MSE of the uniform minimum variance unbiased estimator and the conventional best linear unbiased estimator, respectively. The suitability of the Gibbs Sampler for estimating additional unknown parameters is shown by applying it to the problem in which synchronization of clock drift is also needed.