Search

• Novel control framework based on optimization in the latent space of a SMB-PLS model. • Real-time optimization of continuous process facing raw material fluctuation. • Overcomes parametric disturbances with model adaptation and closed-loop structure. • Adjustment of process operating conditions based on quality thresholds and economics. Reducing the impact of lot-to-lot raw material variability through optimization of operating conditions is key when the lots are already purchased, and available in inventory. The objective of this paper is to provide a framework to optimize operating conditions to maximize profitability while aiming at achieving product quality targets each time a new lot of raw material is fed to a continuous process. The proposed approach consists of solving an optimization problem in the latent space of a sequential multi-block partial least square model (SMB-PLS). Model updating and closed-loop operation are considered to overcome parametric disturbances. The approach is illustrated using a simulated grinding-flotation plant for a sequence of ore lots with variable properties. The case study shows that optimizing operating conditions with the proposed approach allows increasing biannual gain by 1.5 to 2 % compared to nominal operation. This represents between 59 and 75 % of the true achievable gain. [ABSTRACT FROM AUTHOR]

Decision tools to determine the acceptability of an incoming lot of raw materials prior to its purchase from a supplier is essential to ensure efficient quality control. They allow mitigating the propagation of the raw material variability through the process and, ultimately, in the final product quality. Establishing multivariate specifications regions for raw material properties to achieve high and consistent end product quality using latent variable models is key. However, these specifications are typically designed assuming the process operates at nominal conditions. The objective of this work is to enlarge the raw material acceptance region to accept materials having a wider range of properties, and possibly at a lower cost, while meeting final product quality requirements. Using a sequential multi-block partial least squares model (SMB-PLS), the design of multivariate specification regions is combined with optimization of process operating conditions to better cope with fluctuations in raw material properties. An economic criterion is used to assess whether accepting lower cost materials, at the expense of adjusting plant operation, which incurs additional costs, is economically justified. The approach is illustrated using a grinding-flotation plant simulator where the objective is to assess the profitability of different lots of raw materials (i.e. ores). The impact of measurement noise is also investigated. The case studies show that the raw material processability is improved since the number of correctly accepted lots is increased by 7.3 % in comparison with maintaining nominal process conditions. • Adjusting process conditions widens multivariate specifications for raw materials. • Optimization in the latent space of SMB-PLS model is performed for the first time. • Development of multivariate specifications includes an economic criterion. • A general SMB-PLS prediction equation is provided and demonstrated. [ABSTRACT FROM AUTHOR]

The Sequential Multi-Block Partial Least Squares (SMB-PLS) model inversion is applied for defining analytically the multivariate raw material region providing assurance of quality with a certain confidence level for the critical to quality attributes (CQA). The SMB-PLS algorithm does identify the variation in process conditions uncorrelated with raw material properties and known disturbances, which is crucial to implement an effective process control system attenuating most raw material variations. This allows expanding the specification region and, hence, one may potentially be able to accept lower cost raw materials that will yield products with perfectly satisfactory quality properties. The methodology can be used with historical/happenstance data, typical in Industry 4.0. This is illustrated using simulated data from an industrial case study. • Definition of the multivariate raw material specification region from historical data. • Identifying variations in the process uncorrelated with raw materials and disturbances. • Expanding the raw material specification region by modifying the process conditions. [ABSTRACT FROM AUTHOR]

• New dynamic macro scale model of a tablet pan coater based on energy and mass transfer equations. • Modeling for control system design and the development of system observers to estimate critical unmeasured variables in real-time. • Model parametrization and calibration using pilot plant batch data. • Model captures the main process variable trajectories, with errors below 2 °C for the outlet gas and tablet bed temperatures, and 2% for the outlet gas relative humidity. Aqueous film coating is a common and important step in the production of most pharmaceutical tablets. Controlling this process is beneficial to maintain stable and optimal operation and the use of a model capable of describing its main physical phenomena is fundamental for control system design. This paper presents the development of a dynamic macro scale model of a pan coater based on energy and mass transfer equations. Model complexity is kept low to simplify parameterization and allow for real-time applications, such as the development of system observers to estimate critical unmeasured variables. Pilot plant data from seven batches, ran according to designed experiments, are used for model calibration and validation. The final model adequately captures the global trajectories of the main process variables; mean differences between experimental and simulated values are below 2 °C for the outlet gas and tablet bed temperatures, and 2% for the outlet gas relative humidity. [ABSTRACT FROM AUTHOR]

• Individual anode current signals used to detect anode incidents in real-time. • Detection systems built with principal component analysis and an industrial dataset. • Best detection performance was achieved with the simplest system configuration. • Model profit metric developed to evaluate economic potential of detection systems. • Detection systems appear to be economically superior to standard detection method. Anodic incidents occur when an aluminum electrolysis cell short circuits at an anode position. They reduce the current efficiency of the affected cell, which makes it economically advantageous to detect and remove them as early as possible. This study evaluates the performance of three anodic incident detection systems of varying complexity. Individual anode current signals from 6 different cells were monitored with a Principal Component Analysis (PCA) model to detect anodic incidents using the Squared Prediction Error (SPE) statistic. The results suggest that a simple detection system using a common PCA model and a common detection threshold for all cells may be preferable. A novel approach to estimating the economic benefit of a detection system was also developed and used to optimize the systems' detection threshold. The anodic incident detection systems were found to deliver positive profits relative to the standard detection technique. [ABSTRACT FROM AUTHOR]

[Display omitted] In this study, an in-line Process Analytical Technology (PAT) for cosmetic (non-functional) coating unit operations is developed using images of the tablet bed acquired in real-time by an inexpensive industrial camera and lighting system. The cosmetic end-point of multiple batches, run under different operating conditions, is automatically computed from these images using a Multivariate Image Analysis (MIA) methodology in conjunction with a stability determination strategy. The end-points detected by the algorithm differed, on average, by 3% in terms of total batch time from those identified visually by a trained operator. Since traditional practice typically relies on a coating overage to ensure full batch aspect homogeneity in the face of disturbances, the current in-line method can be used to reduce coating material and processing time (over 40% for the operating policy adopted in this work). Additionally, monitoring of the color features calculated by the algorithm allowed the identification of abnormal process conditions affecting visible coating uniformity. This work also addresses practical challenges related to image acquisition in the harsh environment of a pan coater, bringing this tool closer to a state of maturity for implementation in production units and opening the path for their optimization, monitoring, and automatic control. [ABSTRACT FROM AUTHOR]

The Sequential Multi-block PLS algorithm, called SMB-PLS, was recently proposed to improve interpretability of large multi-block data structures. It combines the strengths of Multi-block PLS (MB-PLS) and those of the Sequential Orthogonal PLS (SO-PLS) methods. It uses the two-level hierarchical structure of the first (i.e., block and super levels) providing two levels of scrutiny for the analysis of large datasets, and the sequential orthogonalization scheme of SO-PLS, while keeping between block correlated information in the model. This enables the exploration and interpretation of the full data structure without loss of information. SMB-PLS also allows the selection of a different number of latent variables for each regressor block. The modelling performance and interpretation of SMB-PLS were illustrated using two datasets, covering different types of structural relationships between the regressor blocks. SMB-PLS leads to similar predictive performance of response data as MB-PLS and SO-PLS. However, it was shown that SMB-PLS clearly reveals the correlation structure between the regressor blocks, while MB-PLS leads to more ambiguous results. The correlated information between the blocks extracted with SMB-PLS also improves interpretability, for example, by identifying control actions made to attenuate disturbances, such as raw materials variations. Such information cannot be obtained with SO-PLS since it removes between block correlated variations. [ABSTRACT FROM AUTHOR]

This paper reports on the application of an acousto-ultrasonic (AU) scheme for the inspection of industrial-size carbon anode blocks used in the production of primary aluminium by the Hall-Héroult process. A frequency-modulated wave is used to excite the anode blocks at multiple points. The collected attenuated AU signals are decomposed using the Discrete Wavelet Transform (DTW) after which vectors of features are calculated. Principal Component Analysis (PCA) is utilized to cluster the AU responses of the anodes. The approach allows locating cracks in the blocks and the AU features were found sensitive to crack severity. The results are validated using images collected after cutting some anodes. [ABSTRACT FROM AUTHOR]

A fast and non-invasive near-infrared (NIR) imaging system is proposed to characterize biodegradable polylactic acid (PLA) films containing talc, and produced by extrusion film-blowing for packaging applications. The ultimate goal is to use the system on-line for quality control and after a post-processing thermal treatment (annealing). A set of PLA-talc films with different talc contents (0–3% wt.) were produced and annealed under different conditions. NIR images of the films were collected after which the samples were characterized for their physical, mechanical, and gas barrier properties. Multivariate imaging techniques were then applied on the hyperspectral images. The results showed that the films with various talc loadings can be discriminated, and the information contained in the NIR spectra allowed predicting the annealed film crystallinity and O 2 permeability with good accuracy. [ABSTRACT FROM AUTHOR]

The large datasets resulting from operating HVAC&R systems are currently scrutinized to find ways to exploit the useful information that they might contain. In this work, historical data of a centrifugal water chiller over the course of more than 1.5 years of operation is used to learn about the system and to suggest modifications to its operation scheme. The results show that principal component analysis (PCA) captures well the variance in the historical data. The first two principal components explained between 62 and 80% of the variance, depending on the cases considered. The main factors responsible for the variation of the chiller operation are found to be the weather and the cold water temperature setpoint. The effect of the sampling time step on the results is also studied. Moreover, this work demonstrates that partial least squares (PLS) regression can adequately predict an important indicator of the chiller performance, namely the coefficient of performance (COP), one time step ahead with an R² of 77.49% and root-mean square error of estimation (RMSEE) of 0.463 using a separate validation set of data. The PLS model was also able to predict future COP values up to 2 time steps (∼3 h) in advance. [ABSTRACT FROM AUTHOR]

A novel methodology is proposed for defining multivariate raw material specifications providing assurance of quality with a certain confidence level for the critical to quality attributes (CQA) of the manufactured product. The capability of the raw material batches of producing final product with CQAs within specifications is estimated before producing a single unit of the product, and, therefore, can be used as a decision making tool to accept or reject any new supplier raw material batch. The method is based on Partial Least Squares (PLS) model inversion taking into account the prediction uncertainty and can be used with historical/happenstance data, typical in Industry 4.0. The methodology is illustrated using data from three real industrial processes. • A novel approach to define multivariate specifications overcoming the drawbacks of setting univariate specifications. • Definition of the multivariate raw material region from historical data typical from Industry 4.0 environment. • New methodology to back-propagate the uncertainty in Latent Variable Model Inversion when defining the Design Space. • Evaluation of the capability of a raw material batch prior to introducing it into the production process. [ABSTRACT FROM AUTHOR]

Combining color (spectral) and textural information often leads to improved performance of image-based sensors developed for different applications, such as process monitoring and prediction of process or product quality variables. A new method based on the undecimated wavelet transform and multivariate image analysis (UWT-MIA) is proposed in this paper for simultaneous extraction of spatial and spectral information. The key advantages of this approach are: 1) it uses continuous wavelets which have a constant resolution over all scales compared to discrete wavelets, and are better suited for extracting size distribution features, 2) the wavelet detail and approximation sub-images have the same size and are spatially congruent which allows stacking the sub-images obtained at all scales and orientations (and wavelengths for multi- and hyperspectral images) as a separate channel in a new multivariate image from which a single MIA model can be built. The performance of UWT-MIA is illustrated using both synthetic and natural images. It is shown that variations in color, size and orientation of objects of interest can be tracked efficiently within the same MIA model. The method can also be cast within the multiresolutional multivariate image analysis (MR-MIA) framework proposed in the literature. [ABSTRACT FROM AUTHOR]

To improve understanding of the mixing phenomena resulting from the interaction between uniform magnetic fields and magnetic nanofluids, this work presents an optical investigation of the coherent structures generated by rotating magnetic nanoparticles in dilute nanofluid suspension submitted to uniform rotating magnetic fields. A microscopic visualization technique was implemented inside the bore of an electromagnet to allow image and video capture of the variation in nanofluid behavior under time-varying magnetic fields and different operating parameters. The effects of magnetic core volume fraction, magnetic nanoparticle diameter, rotating magnetic field intensity and frequency as well as magnetic field type on coherent structure generation were explored. The observational results were subsequently correlated and analyzed making use of an automatic structure detection protocol at different scales based on multi-resolution orthogonal 2-Dimensional discrete wavelet transforms. All the parameters considered were found to substantially affect detectable structure inception in the nanofluid in different ways, with trends and thresholds emerging for different types of organisations at different scales. This information now opens the door for adequate tuning of rotating magnetic nanoparticle based mixing technologies at the micro-scale. [ABSTRACT FROM AUTHOR]

The potential and effectiveness of near-infrared (NIR) hyperspectral imaging was investigated for rapid and nondestructive spectral-spatial inspection of functional multilayer films. A series of film samples having different combinations of layers were used to demonstrate the sensor performance. Multivariate Image Analysis (MIA) was applied to NIR hyperspectral images of unprinted multilayer films with different compositions to distinguish them regarding absence/presence of particular layers. Since printing of commercial brands on the films may interfere with product inspection, the ability of MIA to detect the presence of the functional layer (barrier) in spite of film surface printing was also assessed. The results show that the proposed MIA approach is successful at classifying multilayer films, even in the presence of printing. [ABSTRACT FROM AUTHOR]

Undesired particulate agglomeration can create powder flow issues during manufacturing. Potassium chloride (KCl), a commercial product used in powder form as a potassium source, is known to agglomerate. The main objective is to develop a method to assess KCl agglomeration, then use it to understand its driving forces and the phenomena taking place. Based on industrial data in pharmaceutical manufacturing, the following 5 parameters are considered as critical in terms of agglomeration: conditioning humidity, conditioning time, drying, applied pressure and particle size. Beakers containing 40 g of original or ground KCl powder are compacted under specific humidity conditions in a bell jar. Once the beakers are conditioned, agglomerate hardness tests evaluate agglomeration extent by correlating it with the energy required for powder penetration. This energy is calculated from force-distance curves. The results show that the Area Under the force vs distance Curve (AUC) is a good indicator of agglomeration extent. Thus, the AUC is a scalar and has units of work. Based on AUC analysis, the highest agglomeration is found in conditioning humidity of 85% relative humidity (RH), original particle size, drying and conditioning time of 16 h. The agglomeration of original versus ground particle size powders is further investigated over time in a conditioning test. Preliminary tests validate our method and indicate that KCl premixing with other materials can solve agglomeration problems encountered during manufacturing processes as KCl-particle-particle interactions per unit volume are reduced. [ABSTRACT FROM AUTHOR]

The petrography and mineral chemistry of magnetite from fifteen volcanogenic massive sulfide (VMS) deposits in Canada, and the Lasail VMS deposit in Oman, as well as from two VMS-associated banded iron formations (BIF), Austin Brook (New Brunswick, Canada) and Izok Lake (Nunavut, Canada), were investigated using optical microscopy, electron probe micro-analyzer (EPMA), and laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS). The method of robust estimation for compositional data (robCompositions) was applied to investigate geochemical censored data. Among thirty-seven elements analyzed by EPMA and/or LA-ICP-MS in magnetite from the studied deposits/bedrock lithologies, only the results for Si, Ca, Zr, Al, Mg, Ti, Zn, Co and Ni contain < 40% censored values, and thus could be imputed using robCompositions. Imputed censored data were transformed using centered log-ratios to overcome the closure effect on compositional data. Transformed data were classified by partial least squares-discriminant analysis (PLS-DA) to identify different compositional characteristics of magnetite from VMS deposits and BIFs. The integration of petrography and mineral chemistry identifies three types of magnetite in VMS settings: magmatic, hydrothermal, and metamorphic. Magmatic magnetite in VMS deposit host bedrocks is characterized by ilmenite exsolution and may be overprinted by metamorphism. Some VMS deposits contain hydrothermal magnetite, which is intergrown with sulfides, and shows a metamorphic overprint as it is partly replaced by common metamorphic minerals including chlorite, sericite, anthophyllite, and/or actinolite, whereas the majority of the deposits are characterized by metamorphic magnetite formed by replacing pre-existing sulfides and/or silicates, and is intergrown with metamorphic minerals. Among VMS deposits of the Noranda mining district, the West Ansil deposit is characterized by hydrothermal-metamorphic magnetite zoned by inclusion-free cores and Si- and Mg-rich rims. Magnetite from the studied VMS-associated BIFs is also metamorphic in origin. Aluminum, Ti and Zn contents of magnetite can separate BIF from the other mineralized and un-mineralized bedrock lithologies in the studied VMS settings. PLS-DA shows that variable compositions of magnetite slightly discriminate different studied deposits/bedrock lithologies. The geochemical observations suggest that the variation in magnetite chemistry from different VMS settings might be sourced from differences in: 1) the composition and temperature of parental magmas or hydrothermal fluids, 2) the composition of host bedrocks, 3) the composition of co-forming minerals, and 4) oxygen fugacity. PLS-DA distinguishes magnetite compositions from the studied VMS deposits and BIFs from that of the other ore deposit types including Ni–Cu, porphyry Cu-Mo-Au, iron oxide-copper- gold, iron oxide-apatite, and the Bayan Obo REE-Fe-Nb deposit. Magnetite from the VMS settings on average contains lower concentrations of Si, Zr, Al, Mg, Ti, Zn, Co and Ni relative to that from the other mineral deposit types. PLS-DA of magnetite data from VMS deposits and BIFs of the Bathurst mining camp as well as PLS-DA of magnetite compositions from various mineral deposit types yield discrimination models for application to mineral exploration for VMS deposits using indicator minerals in Quaternary lithified sedimentary rocks. [ABSTRACT FROM AUTHOR]

Magnetite grains from the Izok Lake (Nunavut, Canada) and the Halfmile Lake (New Brunswick, Canada) volcanogenic massive sulfide deposits, and from till covering the nearby areas were investigated using the scanning electron microscopy (SEM), electron probe micro-analyzer (EPMA), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), and optical microscopy. The method of robust estimation for compositional data (rob-composition) was applied to censored geochemical data, and the results were analyzed by principal component analysis (PCA). Textural relationships and mineral association of magnetite reveal the history of formation, and contribute to the explanation of characteristic compositional differences of magnetite from different geological settings. The integration of petrography and mineral chemistry allows discriminating magmatic, metamorphic and hydrothermal magnetite grains in the VMS deposits bedrock samples. Magmatic magnetite is found in Izok Lake gabbro, and Halfmile Lake syenite, felsic ash tuff and gossan samples, whereas magnetite in Izok Lake massive sulfides, gahnite-rich dacite and iron formations formed during the amphibolite facies metamorphism. In Halfmile Lake andesite, magnetite recrystallized during greenschist facies metamorphism. In the magnetite alteration zone associated to the Halfmile Lake deposit, hydrothermal magnetite has been overprinted by metamorphic magnetite. Halfmile Lake massive sulfides in chloritic argillite contain hydrothermal magnetite. PCA identifies discriminator elements and their contributions to magnetite composition from different Izok Lake Lake and Halfmile Lake bedrock samples. The results suggest that Si, Ca, Zr, Al, Ga, Mn, Mg, Ti, Zn, Co, Ni and Cr are discriminator elements for VMS deposits and their host bedrocks. The distinct chemical signatures for magnetite from various bedrock lithologies demonstrate that magnetite grains of the same origin share more similarities in chemistry, as high Ti indicates magmatic sources for magnetite, whereas high Si, Ca and Mg are indicative of hydrothermal settings. Variable compositions of metamorphic magnetite suggest that the chemistry of this type of magnetite is controlled by the composition of host rocks, the grade of metamorphism and oxygen fugacity. PCA of EPMA and LA-ICP-MS data of magnetite from the Izok Lake and Halfmile Lake bedrock samples yield discrimination models for classification of magnetite grains from till. Decreases in the proportion of magnetite grains with the chemical signature of the Izok Lake massive sulfides and gahnite-rich dacite down-ice from the Izok Lake deposit show the use of magnetite chemistry in geochemical exploration. In the Halfmile Lake area, till magnetite grains with the signature of VMS mineralization make a glacial dispersal train more than 2 km down-ice from the deposit. [ABSTRACT FROM AUTHOR]

The estimation of the flotation rate constant is generally considered difficult because of the number of variables – not always measurable – determining its value. Among them, the particle size distribution and hydrodynamic characteristics are considered key elements. Part 1 of this paper introduced the interfacial area of bubbles ( I b ) as a hydrodynamic variable providing more information about the size distribution than the bubble surface area flux ( S b ). Fundamental expressions were proposed to characterize I b using the population mean and standard deviation. Experimental results indicated that for lognormal bubble size distributions, I b correlates very well with the gas hold-up and d 32 . Part 2 investigates the correlation between the flotation rate constant and particle size as well as given hydrodynamic variables using a Projection to Latent Structures (PLS) analysis. The tests were conducted under ‘ideal’ conditions (i.e. shallow froth, low mineral concentration and pure mineral particles). Results suggest that for the fine particle sizes, the bubble surface area flux ( S b ) should be considered for the kinetic constant modeling. For coarser particle, the gas hold-up ( ε g ) is the determining parameter. In practice though, the particle size distribution often lies between these two extreme cases, and can either span a very large range or contains intermediate size particles. In such cases, the interfacial area of bubbles ( I b ) better correlates with the flotation kinetics. [ABSTRACT FROM AUTHOR]