Multi-unit Process Research Articles

Capturing connectivity information from process flow diagrams by sequential-orthogonalized PLS to improve soft-sensor performance

In the development of data-driven soft sensors for product quality assessment in multi-unit manufacturing processes, the only information that is typically used as an input to the model is real-time measurements from field sensors. However, even if detailed knowledge of the mechanistic behavior of the process may not be available, information about the sequence of processing units, and their connectivity, is available, typically in graphical form through process flow diagrams. In this study, we investigate the use of sequential-orthogonalized partial least-squares (SO-PLS) regression as a way to capture connectivity information from a process flow diagram, and transfer it into a data-driven model to be used as a soft sensor in a multi-unit process. Connectivity between units is captured and translated into a block order that establishes a sequence for block regressions. Orthogonalization between two blocks is then carried out with the aim of eliminating overlapping data and retaining information that is unique to each block. Product quality is finally predicted by summing the contributions from each block, and the accuracy of prediction is enhanced due to the embedded dual feature-extraction procedure, which combines orthogonalization and latent-variable extraction. The effectiveness of the proposed approach is illustrated by comparing the quality prediction performance of two soft sensors for a simulated multi-unit continuous process: one using standard PLS and one using SO-PLS. Superior performance of the SO-PLS soft sensor is achieved, even more markedly so when fewer field measurements are available to build the soft sensor.

An advanced treatment process for 3-high wastewater discharged from crude oil storage tanks.

The wastewater discharged from crude oil storage tanks (WCOST) contains high concentrations of salt and metal iron ions, and high chemical oxygen demand (COD). It belongs to "3-high" wastewater, which is difficult for purification. In this study, WCOST treatments were comparatively investigated via an advanced pretreatment and the traditional coagulation-microfiltration (CMF) processes. After WCOST was purified through the conventional CMF process, fouling occurred in the microfiltration (MF) membrane, which is rather harmful to the following reverse osmosis (RO) membrane unit, and the effluent featured high COD and UV254 values. The analysis confirmed that the MF fouling was due to the oxidation of ferrous ions, and the high COD and UV254 values were mainly attributable to the organic compounds with small molecular sizes, including aromatic-like and fulvic-like compounds. After the pretreatment of the advanced process consisting of aeration, manganese sand filtration, and activated carbon adsorption in combination with CMF process, the removal efficiencies of organic matter and total iron ions reached 97.3% and 99.8%, respectively. All the water indexes of the effluent, after treatment by the advanced multi-unit process, meet well the corresponding standard. The advanced pretreatment process reported herein displayed a great potential for alleviating the MF membrane fouling and enhanced the lifetime of the RO membrane system in the 3-high WCOST treatment.

Optimized Gaussian-Process-Based Probabilistic Latent Variable Modeling Framework for Distributed Nonlinear Process Monitoring

Plant-wide multiunit processes generally contain numerous variables, complex relations, and strong nonlinearity, making the monitoring of such processes challenging. This work proposes a new Gaussian-process-based probabilistic latent variable (GPPLV) modeling framework for distributed monitoring of multiunit nonlinear processes. A Gaussian-process latent variable model is first established to extract the dominant features of a local unit. Using the extracted features, a correlation between the local unit and its neighboring units are then modeled through a Gaussian-process regression (GPR) model. The genetic algorithm is used to determine the ideal independent variables from the neighboring units and optimize the hyperparameters of the GPR model simultaneously. Residuals are generated and monitoring statistics are constructed using an established GPPLV model. Experimental studies on three processes: 1) a numerical example; 2) the Tennessee Eastman benchmark process; and 3) a laboratory distillation process show that compared to some common distributed process monitoring models, the proposed method performs better in showing the nature of different faults and shows higher fault detection rate for large-scale multiunit processes.

Self-attention-based Multi-block regression fusion Neural Network for quality-related process monitoring

BackgroundFor plant-wide process with multiple operation units, local-global modeling is an efficient method to achieve quality-related fault detection. However, most of algorithms based on local-global modeling ignore the correlation between sub-blocks. This will result in poor performance of the extracted global quality-related features. MethodsThis paper focus on the correlation between sub-blocks and proposes Self-attention-based Multi-block regression fusion Neural Network (SMNN) to achieve efficient quality-related fault detection for nonlinear multi-unit process. Firstly, to focus on quality-related information, the key variables are selected. Then, to extract quality-related features in each sub-block, a pre-training approach is used, i.e. a deep neural network-based regression network between process variables and quality variables is constructed in each sub-block. Secondly, considering the correlation between the sub-blocks, self-attention mechanism is used to integrate the quality-related feature from each block. With the help of an additional regression network, the quality-related features of sub-blocks are fine-tuned and the global features are extracted. Finally, quality-related statistic is constructed to detect faults. FindingsThe proposed method shows good performance in Tennessee-Eastman process, which demonstrates the effectiveness of the method. It also shows that considering the potential relationships between sub-blocks during model construction helps in the extraction of global features.

Variational Bayesian probabilistic modeling framework for data-driven distributed process monitoring

Data-driven process monitoring has gained increasing attention because of the increasing demand in process safety and the rapid advancement of data gathering techniques. When monitoring a plant-wide multiunit process, establishing a monitor for each unit individually ignores the correlations among units, whereas establishing a global monitor for the entire process ignores the local process behavior. A variational Bayesian-based probabilistic modeling approach is proposed for efficient distributed process monitoring. A novel probabilistic latent variable model is developed to characterize the variable relationship in each local unit and among units. First, variational Bayesian-based latent variable extraction is performed in each local unit, through which variable relationship within a local unit is characterized. Second, variational Bayesian-based regression model is established between the latent variables and neighboring variables, through which the variable relationship among units is characterized. Then, modeling residuals and monitoring statistics are generated, through which the process status and the type of a detected fault are identified. The effectiveness of the proposed probabilistic modeling and monitoring method is verified by three case studies, including a numerical example, the Tennessee Eastman benchmark process, and a laboratory distillation process.

A simulation-based optimisation framework for process plan generation in reconfigurable manufacturing systems (RMSs) in an uncertain environment

ABSTRACT Reconfigurable manufacturing system (RMS) is a manufacturing paradigm which is proven to be time and cost-effectively adaptable to a wide range of market changes due to its customisable capacity and functionality. In this paper, a two-step framework is proposed for the process plan generation in RMS. The first step, aimed at solving the multi-objective part-family Single-Unit Process Plan (SUPP) generation problem, involves a comparative approach using three metaheuristics, namely: The Non-Dominated Sorting Genetic Algorithm (NSGA-II), the Archived Multi-Objective Simulated Annealing (AMOSA) and the Multi-Objective Particle Swarm Optimisation (MOPSO) as well as a simulation-based optimisation method. The second step is designed to solve the multi-objective part-family Multi-Unit Process Plan (MUPP) generation problem with unpredictable demands in different periods using a combination of the answers of the algorithms in step 1. The number of units is also optimised using the NSGA-II. Finally, a novel heuristic algorithm named Designed Periods Algorithm (DPA) is proposed in the second step to meet the unpredictable demands in different periods. To illustrate the applicability of the framework, an example is presented, the results of which have shown the superiority of the MUPP over the SUPP in response to unpredictable demands according to the periods designed by DPA.

Distributed-ensemble stacked autoencoder model for non-linear process monitoring

Determining whether a fault occurs locally or globally is highly important for large-scale industrial processes involving multiple operating units. Moreover, the complex non-linearity among process variables is a prominent feature of modern industries. This paper proposes a distributed-ensemble stacked autoencoder (DE-SAE) model based on deep learning technology for monitoring non-linear, large-scale, multi-unit processes. First, the deep features of the variables involved in each operating unit are extracted with the stacked autoencoder (SAE) to represent the essential structure of the unit. Two statistics are separately constructed using the deep features and the reconstruction error for detecting the faults in local units. Subsequently, the deep representations of the variables from each operating unit are modeled with the SAE to extract the global information for global monitoring. The proposed DE-SAE model uses deep learning techniques to solve the complex non-linear relationships in industrial processes, while considering their local and global information. Therefore, the method can explain the monitoring results better. Experimental results obtained from the numerical simulation and Tennessee-Eastman process confirm the feasibility and superiority of this method.

Neighborhood Variational Bayesian Multivariate Analysis for Distributed Process Monitoring With Missing Data

Conventional methods for distributed monitoring commonly assume that complete process measurements are available. However, the problem of missing data is often encountered in the monitoring of large-scale multiunit processes. This paper proposes an approach based on a neighborhood variational Bayesian principal component analysis (NVBPCA) and canonical correlation analysis (CCA) for the efficient distributed monitoring of multiunit processes in the presence of missing data. Missing observations for a local unit are reconstructed through NVBPCA by considering information from both local and neighboring units. A CCA-based local monitor, which identifies the status of the local unit and the type of a detected fault using information from both the local and neighboring units, is then developed. The NVBPCA–CCA approach has a better performance since its missing data handling and local monitor construction consider information from both the local and neighboring units. The efficiency of the proposed monitoring method is demonstrated through its application in a numerical example and an industrial tail gas treatment process.

A Novel Framework to Aid the Development of Design Space across Multi-Unit Operation Pharmaceutical Processes-A Case Study of Panax Notoginseng Saponins Immediate Release Tablet.

The fundamental principle of Quality by Design (QbD) is that the product quality should be designed into the process through an upstream approach, rather than be tested in the downstream. The keystone of QbD is process modeling, and thus, to develop a process control strategy based on the development of design space. Multivariate statistical analysis is a very useful tool to support the implementation of QbD in pharmaceutical process development and manufacturing. Nowadays, pharmaceutical process modeling is mainly focused on one-unit operations and system modeling for the development of design space across multi-unit operations is still limited. In this study, a general procedure that gives a holistic view for understanding and controlling the process settings for the entire manufacturing process was investigated. The proposed framework was tested on the Panax Notoginseng Saponins immediate release tablet (PNS IRT) production process. The critical variables and the critical units acting on the process were identified according to the importance of explaining the variability in the multi-block partial least squares path model. This improved understanding of the process by illustrating how the properties of the raw materials, the process parameters in the wet granulation and the compaction and the intermediate properties affect the tablet properties. Furthermore, the design space was developed to compensate for the variability source from the upstream. The results demonstrated that the proposed framework was an important tool to gain understanding and control the multi-unit operation process.

Multi-objective multi-unit process plan generation in a reconfigurable manufacturing environment: a comparative study of three hybrid metaheuristics

Low costs, high reactivity and high quality products are necessary criteria for industries to achieve competitiveness in nowadays market. In this context, reconfigurable manufacturing systems (RMSs) have emerged to fulfil these requirements. RMS is one of the latest manufacturing paradigms, where machines components, software or material handling units can be added, removed, modified or interchanged as needed and when imposed by the necessity to react and respond rapidly and cost-effectively to changing. This research work addresses the multi-objective single-product multi-unit process plan generation problem in a reconfigurable manufacturing environment where three hybrid heuristics are proposed and compared namely: repetitive single-unit process plan heuristic (RSUPP), iterated local search on single-unit process plans heuristic (LSSUPP) and archive-based iterated local search heuristic (ABILS). Single-unit process plans are generated using the adapted non-dominated sorting genetic algorithm (NSGA-II). Moreover, in addition to the minimisation of the classical total production cost and the total completion time, the minimisation of the maximum machines exploitation time is considered as a novel optimisation criterion, in order to have high quality products. To illustrate the applicability of the three approaches, examples are presented and the obtained numerical results are analysed.

Optimal Variable Transmission for Distributed Local Fault Detection Incorporating RA and Evolutionary Optimization

Determining the variable transmission structure is the key step in designing a distributed monitoring scheme for multiunit processes. This paper proposes randomized algorithm (RA) integrated with evolutionary optimization-based data-driven distributed local fault detection scheme to achieve efficient monitoring of multiunit chemical processes. First, the RA is employed to generate faulty validation data. Second, evolutionary optimization-based variable transmission structure determination is performed to achieve the minimal non-detection rate by selecting transferred variables. Then, a principal component analysis (PCA) or kernel PCA monitoring model is established for each operation unit to identify the status of the unit. Last, a comprehensive index is developed to identify the status of the entire process. The established local monitors consider the relationship with other units but avoid introducing redundant information, thereby exhibiting superior monitoring performance. Case studies on two numerical examples, including a linear and a nonlinear case and the Tennessee Eastman benchmark process, are provided. Comparative results of traditional local or global monitoring methods verify the efficiency of the proposed monitoring scheme.

Quantification of irreversibilities in practical cyclic processes using exergy analysis and Gouy-Stodola theorem

The exergy analysis of a process to quantify the irreversibilities is advantageous over the entropy analysis in that it provides the definition of efficiency of a process, referred to as exergetic efficiency (defined as the ratio of exergy recovered to exergy supplied to the process). Unfortunately, the exergy analysis of practical multi-unit cyclic processes is rarely covered adequately in the undergraduate courses in engineering thermodynamics. In this article, the quantification of irreversibilities is illustrated in detail for a practical cyclic steam power plant using exergy analysis and Gouy-Stodola theorem. The efficiencies are determined for the various components and for the whole process. The theoretical background related to exergy, exergy analysis, and Gouy-Stodola theorem is also covered briefly for the benefit of the students. An assessment problem dealing with vapor-compression refrigeration cycle is included at the end in order to assess the intended learning outcomes of this article. The key solution steps along with answers are also provided for the benefit of the readers. As exergy analysis involves advanced level concepts in thermodynamics, the appropriate place for the introduction of the material presented in this article is the second, advanced level, course in thermodynamics.

Statistical modeling methods to analyze the impacts of multiunit process variability on critical quality attributes of Chinese herbal medicine tablets.

The quality of Chinese herbal medicine tablets suffers from batch-to-batch variability due to a lack of manufacturing process understanding. In this paper, the Panax notoginseng saponins (PNS) immediate release tablet was taken as the research subject. By defining the dissolution of five active pharmaceutical ingredients and the tablet tensile strength as critical quality attributes (CQAs), influences of both the manipulated process parameters introduced by an orthogonal experiment design and the intermediate granules’ properties on the CQAs were fully investigated by different chemometric methods, such as the partial least squares, the orthogonal projection to latent structures, and the multiblock partial least squares (MBPLS). By analyzing the loadings plots and variable importance in the projection indexes, the granule particle sizes and the minimal punch tip separation distance in tableting were identified as critical process parameters. Additionally, the MBPLS model suggested that the lubrication time in the final blending was also important in predicting tablet quality attributes. From the calculated block importance in the projection indexes, the tableting unit was confirmed to be the critical process unit of the manufacturing line. The results demonstrated that the combinatorial use of different multivariate modeling methods could help in understanding the complex process relationships as a whole. The output of this study can then be used to define a control strategy to improve the quality of the PNS immediate release tablet.

Design and control of an improved acrylic acid process

The acrylic acid (AA) process involves the partial oxidation of a flammable and explosive gas medium (propylene), so considerable attention is paid to the concentration of the reactants. Water and air is added to dilute the reactant and enhance the thermal stability of the reaction. Different compositions of water and air can lead to different methods to separate AA product. A modified AA process is developed based on the AA process proposed by Turton, and consists of a tubular reactor, an absorber and only two distillation columns, one of which is an azeotropic distillation column. This process is characterized by shortening and simplifying AA refining process so that equipment investment cost can be reduced. A plantwide control structure featured with cascade control provides effective control for the multiunit process and insures safe operation of the two distillation columns. Dynamic results also reveal that it is useful to apply temperature/temperature cascade control and composition/temperature cascade control to the distillation columns with maximum temperature limitations.

PLANTWIDE CONTROL OF A COMPLEX PROCESS INVOLVING A REACTIVE DISTILLATION COLUMN

In this study, several plantwide control structures for a multiunit process including a reactive distillation column are proposed. The process studied contains two reaction and three separation steps, two recycle streams, and six components. The second reaction of the process takes place in a reactive distillation column, while the first reaction occurs in a continuous stirred tank reactor. Flow control of a stream in each recycle loop is required for base-level regulatory control of the process in a reasonable settling time. The quality of the first product is improved using a dual-temperature control in the second distillation column. The product quality in the reactive column is improved using a three-temperature control structure.

A real time approach to process control education with an open-ended simulation tool

Process simulators are widely used in industrial process designs and academic research. These simulation tools are also perfectly suitable for the process dynamics and control education of junior chemical engineers and students, as these tools mimetically help them with comprehending the basic theories of process control, such as process capacity, dead time, control loops, controllers and multi-unit processes such as distillation columns. At the University of Auckland, New Zealand, final year Chemical and Materials Engineering students who participate in the process dynamics and control paper are required to complete a series of simulation workshops in auxiliary sessions to help them in their understanding of process dynamics and control. This paper introduces the content of the workshops as well as reviews the student feedback on the introduction of the simulator and their perceptions of their learning of process dynamics and control as assisted by the software and instruction. Three case studies are provided in this paper to illustrate the benefits of running workshops. The motivation of this paper is to share our workshop design with other universities.

Guiding Principles for Accounting Consumption of Energy in Multi-unit Process Industries

ABSTRACT Most manufacturing processes involve many heating and cooling phase transformation cycles. The end product(s) often involves inputs from more than one process or different stages of processing, and in each stage or process there are many energy inputs and outputs apart from the product itself. Furthermore, there could be stocking and destocking of intermediate product(s). All these lead to a situation wherein computation and comparison of specific energy consumption as an index of efficiency and cost of operation becomes impossible. In this article, a methodology is presented that permits computation of specific energy consumption for the individual processes/stages, as well as the overall balanced specific energy consumption for the entire process.

A two‐level distillation design method

AbstractRecently, Lucia et al. have used a distillation line method to develop the concept of shortest stripping line distance approach to minimum energy designs of distillation columns and multiunit processes. It is well known that distillation line methods can be very sensitive to specified product compositions. A two‐level distillation design procedure is proposed for finding portfolios of minimum energy designs when specifications are given in terms of key component recoveries. Thus, product compositions are not specified but calculated. It is shown that the proposed two‐level design procedure is flexible and can find minimum energy designs for both zeotropic and azeotropic distillations. It is also shown that the two‐level design method encompasses Underwood's solution but can find minimum energy designs when Underwood's method fails. Numerical results for several distillation examples involving ternary and quaternary mixtures are presented to support these claims, and geometric illustrations are used to elucidate key points. © 2008 American Institute of Chemical Engineers AIChE J, 2008

Non-pinched, minimum energy distillation designs

Non-pinched, minimum energy solutions are important class of distillation designs that offer the potential advantage of a better trade-off between capital investment and operating costs. In this paper, two important tasks associated with non-pinched distillation designs are studied. Thus the novel contributions of this work to the literature are (1) A comprehensive methodology for finding non-pinched minimum energy designs. (2) Understanding of the reasons for the existence of non-pinched distillation designs. It is shown that the recent shortest stripping line distance approach of Lucia et al. [Lucia, A., Amale, A. and Taylor, R., 2007, Distillation pinch points and more. Comput Chem Eng, available on-line] is capable of systematically and reliably finding non-pinched, minimum energy distillation designs. In addition, we provide an understanding of the reasons behind the existence of non-pinched designs, which include trajectories that follow unstable branches of a pinch point curve in azeotropic systems, the inherent looping structure of trajectories in hydrocarbon separations, and the presence of ancillary constraints in multi-unit processes like extraction/distillation. Several distillation examples are studied and many numerical results and geometric illustrations are presented that show the shortest stripping line distance methodology is indeed a powerful and systematic tool for computing non-pinched, minimum energy designs and that support the underlying reason we provide for the existence of non-pinched designs.

Quantitative comparison of dynamic controllability between a reactive distillation column and a conventional multi-unit process

A comparison of the steady-state economic optimum designs of two alternative chemical processes was presented in a previous paper [Kaymak, D. B., & Luyben, W. L. (2004). A quantitative comparison of reactive distillation with conventional multi-unit reactor/column/recycle systems for different chemical equilibrium constants. Industrial & Engineering Chemistry Research, 43, 2493–2507]. A generic exothermic reversible reaction A + B ↔ C + D occurs in both flowsheets, which consist of a conventional multi-unit reactor/separator/recycle structure and a reactive distillation column. Results showed that the reactive distillation process is significantly less expensive than the conventional process for a wide range of the chemical equilibrium constant when there is no mismatch between the temperature favorable for reaction and the temperature favorable for vapor–liquid separation. A reactive distillation column has fewer control degrees of freedom than a conventional multi-unit system. Therefore a reactive distillation column may have worse dynamic response than a conventional process. The purpose of this paper is to compare the dynamic controllability of these two alternative processes. Three different chemical equilibrium constants are considered. Several control structures are developed for each flowsheet, and their effectiveness is evaluated. Disturbances in production rate and fresh feed compositions are considered. The conventional multi-unit process provides significantly better control. The operability region is much larger, there is less variability in product quality and the dynamic responses are faster than those of the reactive column. Thus, these results demonstrate that there is a significant trade-off in this system between optimum economic steady-state design and dynamic controllability.