Comparison of linear and nonlinear dimension reduction techniques for automated process monitoring of a decentralized wastewater treatment facility

Multivariate statistical methods for online process monitoring have been widely applied to chemical, biological, and engineered systems. While methods based on principal component analysis (PCA) are popular, more recently kernel PCA (KPCA) and locally linear embedding (LLE) have been utilized to better model nonlinear process data. Additionally, various forms of dynamic and adaptive monitoring schemes have been proposed to address time-varying features in these processes. In this analysis, we extend a common simulation study in order to account for autocorrelation and nonstationarity in process data and comprehensively compare the monitoring performances of static, dynamic, adaptive, and adaptive–dynamic versions of PCA, KPCA, and LLE. Furthermore, we evaluate a nonparametric method to set thresholds for monitoring statistics and compare results with the standard parametric approaches. We then apply these methods to real-world data collected from a decentralized wastewater treatment system during normal and abnormal operations. From the simulation study, adaptive–dynamic versions of all three methods generally improve results when the process is autocorrelated and nonstationary. In the case study, adaptive–dynamic versions of PCA, KPCA, and LLE all flag a strong system fault, but nonparametric thresholds considerably reduce the number of false alarms for all three methods under normal operating conditions.     

Vitriclastic lithofacies from Macquarie Island (Southern Ocean): compositional influence on abyssal eruption explosivity in a dying Miocene spreading ridge

Macquarie Island is composed of a complete section of oceanic crust that formed in a slow-spreading mid-ocean ridge 2.0 to 3.5 km below sea level. Vitriclastic facies preserved on the island have both pyroclastic and hyaloclastic characteristics. Monomict hyaloclastic breccia facies are widespread across the island and are predominantly composed of near-primitive (~7.9 wt% MgO) subalkaline/transitional (~0.7 wt% K₂O) sideromelane shards and crystalline basalt clasts with low vesicularity (LV, < 15% vesicles). Breccias are thick bedded and structureless with matrix-supported angular pillow fragments, bomb-sized fluidal mini-pillows, and globular glass lapilli. Clasts are lithologically similar to interbedded pillow basalts and laterally grade into fine-grained sandstone facies. These sandstones are normal-graded, well-laminated, thin bedded, and interstratified with red pelagic mudstone. Lithofacies associations indicate that the hyaloclastic breccias were formed proximal to a source vent via quench-fragmentation, and subsequently reworked by ocean-bottom currents into distal epiclastic sandstone facies. During eruption, co-genetic pillow lava and hypabyssal intrusions mingled with the breccia, forming fluidal peperite. Rare polymict pyroclastic facies only occur in the highest stratigraphic levels and are mostly composed of highly vesicular (HV, 15–50% vesicles) sideromelane shards and crystalline basalt clasts with alkaline (~1.0 wt% K₂O) fractionated (~6.8% MgO) compositions. Minor lithic grains are composed of subalkaline (~0.7 wt% K₂O) to very highly alkaline (~1.7 wt% K₂O) LV sideromelane shards, and amphibole-bearing diabase. The pyroclastic facies contains medium to thick beds of lapilli-tuff that exhibit both reverse and normal grading, diffuse lamination, and planar-grain fabric. These beds are locally overlain by thin fine-grained tuff beds entirely composed of cuspate to very thin elongate bubble-wall shards. These characteristics indicate that explosive deep-marine eruptions produced high-density coarse-grained gravity flows that were covered by slower suspension settle-out of delicate bubble-wall shards. Stratigraphic relationships suggest that explosive eruptions started during the waning stages of more alkaline volcanism along the proto-Macquarie spreading center.     

Shear wave velocities of Mississippi embayment soils from low frequency surface wave measurements

Deep unconsolidated sediments in the Mississippi embayment will influence ground motions from earthquakes in the New Madrid seismic zone. Shear wave velocity profiles of these sediments are important input parameters for modeling wave propagation and site response in this region. Low-frequency, active-source surface wave velocity measurements were performed to develop small-strain shear wave velocity (V_S) profiles at eleven deep soil sites in the Mississippi embayment, from north of New Madrid, Missouri to Memphis, Tennessee. A servo-hydraulic, low-frequency source was used to excite surface wave energy to wavelengths of 600 m, resulting in V_S profiles to depths of over 200 m. The average V_S profile calculated from the eleven sites is in good agreement with common reference V_S profiles that have been used in seismic hazard studies of this region. The variability in V_S profiles is shown to be associated with changes in formation depth and thickness from site-to-site. Using lithologic information at each site, average formation velocities were developed and compared to previous studies. We found average V_S values of about 193 m/s for alluvial deposits, 400 m/s for the Upper Claiborne formations, and 685 m/s for the Memphis Sand formation.     

A compact representation of gamma-ray burst time series

Multiscale edge detection (MSED), using wavelet transform extrema, provides a robust method to compress the information in a transient signal. We apply this to gamma-ray burst (GRB) time series. Multiscale edge detection can be used to quantify the variability, identify structures (e.g. FREDs), and suppress noise. We present preliminary results of MSED applied to BATSE 64ms discsc data.     

Modeling outflow from the Ernest Henry Fe oxide Cu–Au deposit: implications for ore genesis and exploration

The Ernest Henry Fe oxide Cu–Au (IOCG) deposit (>ca. 1.51 Ga) is hosted by breccia produced during the waning stages of an evolving hydrothermal system that formed a number of tens of metres to a kilometre scale, pre- and syn-ore alteration halos, although no demonstrable patterns have been attributed to fluids expelled through the outflow zones. However, the recognition of a population of hypersaline fluid inclusions representing the ‘spent’ fluids after Cu–Au deposition at Ernest Henry provides the basis to model the geochemical characteristics of the deposit's outflow zones. Geochemical modeling at 300 °C was undertaken at both high and low fluid/rock ratios via FLUSH models involving three host rock types: (1) granite, (2) calc–silicate rock, and (3) graphitic schist. In models run at high fluid/rock ratios, all rock types are essentially fluid-buffered, and produce an albite–quartz–hematite–barite-rich assemblage, although in low fluid–rock environments, the pH, redox, and geochemical character of the host rock exerts a greater influence on the mineralogy of the alteration assemblages (e.g., andradite, Fe–chlorite, and magnetite). Significant sulphide mineralization was predicted in graphitic schist where sphalerite occurred in both low- and high-porosity models, which indicates the possibility of an association between high-temperature IOCG mineralization and lower temperature base metal mineralization.Cooling experiments (from 300 to 100 °C) using the ‘spent fluids’ predict early high-T (300–200 °C) Na-, Ca-, Fe-, and Mn-rich, magnetite-bearing hydrothermal associations, whereas with cooling to below 200 °C, and with progressive fluid–rock interaction, the system produces rhodochrosite-bearing, hematite–quartz–muscovite–barite-rich assemblages. These results show that the radical geochemical and mineralogical changes associated with cooling and progressive fluid influx are likely to be accompanied by major transformations in the geophysical expression (e.g., spectral and magnetic character) of the alteration in the outflow zone, and highlight the potential link between magnetite- and hematite-bearing IOCG hydrothermal systems.     

Review of the tectonics of the Levant Rift system: the structural significance of oblique continental breakup

The Levant Rift system is an elongated series of structural basins that extends for more than 1000 km from the northern Red Sea to southern Anatolia. The system consists of three major segments, the Jordan Rift in the south, El Gharb–Kara-Su Rift in the north, and the Lebanese Fault splay in between. The rifted parts of this structural system are accompanied by intensively uplifted margins that mirror-image the basinal pattern, namely, the deeper the basin—the higher its margins, and vice versa. Uplifts also occur along the fault splay section. The Jordan Rift comprises axial basins that diminish in size from the south northwards, and are separated from each other by shallow threshold zones along the axis of the rift, where the margins are also subdued. The Lebanese Fault splay consists of five faults that emerge from the northern edge of the Jordan Rift and trend like a fan between the north and the northeast. One of these faults connects the Jordan and El Gharb–Kara-Su rifts. The Levant Rift and its uplifted margins started to develop in the middle-late Miocene, and most of the structural development occurred in the Plio-Pleistocene.The Levant Rift system is characterized by its oblique displacement, and evidence for both dip-slip and strike-slip displacement was measured on its faults. Earthquakes also indicate that same mixed pattern, some of them show strike-slip offset, and others normal. It is generally conceded that the amount of normal offset along the boundary faults of the Rift system reaches 8–10 km, but the lateral displacement is disputed, and offsets ranging from 11 to 107 km were suggested. Assessment of the available data led us to suggest that the sinistral offset along the Levant Rift system is approximately 10–20 km. The similarity between the vertical and the lateral displacements, the basin and threshold structural pattern of the Rift, model experiments in oblique rifting, as well as the significant tectonic resemblance to the Red Sea and the East African rifts, indicate that the Levant Rift is the product of continental breakup, and it is probably an emerging oceanic spreading center.     

Automated material balance and assay data adjustment around a piece of mineral processing equipment

This paper describes an algorithm for calculating the minimum adjustment required to make a set of assays, measured around a single piece of processing equipment, consistent with the mass flows present in that system. This algorithm is then applied to the problem of hydrocyclone size data reconciliation.     

Natural and synthetic leucites studied by solid state 29-Si and 27-Al NMR and 57-Fe Mossbauer spectroscopy

The 29-Si NMR spectra of natural and synthetic leucites (KAlSi₂O₆) are found to contain a number of resonances which are interpreted in terms of the known structure of low-temperature (tetragonal) leucite. Computer simulation of the spectra suggests that the most distorted tetrahedral lattice site T ₁ contains a higher proportion of Al than the other two tetrahedral sites. The occurrence of some ordering of the tetrahedral Si and Al in leucite is confirmed by Mossbauer studies of synthetic iron-containing leucites, including the fully ferric end-member KFeSi₂O₆, in which the three tetrahedral sites can be distinguished. On replacement of about half this Fe by Al, the most distorted of these sites is lost from the spectrum, reflecting the preference of Al for this site. A linear relationship is found between the unit cell dimensions of all these leucites and their iron content.     

A reappraisal of the Rb-Sr systematics of early Archaean gneisses from Hebron,Labrador

A re-investigation of the Rb-Sr isotope systematics of early Archaean granodiorite gneisses from Hebron, northern Labrador, confirms the existence of early Archaean crust in that area, but has failed to corroborate both the high degree of coherence and the high initial⁸⁷Sr/⁸⁶Sr reported by Barton [1]. Instead, marked geological scatter is observed in the data for both slabbed gneisses and large bulk samples, a scatter unequivocally due to the development of secondary whole-rock isochrons at ～ 1800 Ma.Regression analysis of the new data gives very large uncertainties in the age and initial⁸⁷Sr/⁸⁶Sr because of this secondary disturbance, viz. 3645_?470⁺⁸⁷⁵ Ma and 0.702_+0.005^?0.009. We interpret the previous indications of high initial⁸⁷Sr/⁸⁶Sr at 3600 Ma as due to chance.The Hebron data cannot be distinguished from similar populations of geologically-disturbed Rb-Sr results from the Uivak I gneisses to the north [5] and the Amîtsoq grey gneisses [6]. This supports a field-based contention that the so-called “Hebron gneisses” are both lithologic and stratigraphic equivalents of the Uivak I gneisses.