Real Plant Data Research Articles

Development of advanced core noise monitoring system for BWRS

A BWR core noise monitoring system is developed for addressing core anomaly problems in future advanced core operation. In order to monitor in-core status from a limited number of signals, various up-to-date signal processing algorithms are introduced to compensate for a lack of information. These algorithms, such as independent component analysis, factor analysis and model based parameter estimation, are demonstrated to be effective through real plant data analysis to evaluate core and regional stability index, reactivity coefficients and core flow rate. Through these practices, we demonstrate that the core noise monitoring system is an effective general platform for providing a variety of monitoring tools to meet the requirements in future advanced core operation.

Overall Statistical Monitoring of Static and Dynamic Patterns

A new approach to overall multivariate statistical monitoring, including all static normal operations and the intermediate states between them, is introduced and then applied to real plant data. Principal component analysis or partial least squares (PLS) is used to reduce the dimensionality of data and to remove collinearity. After the compression of data, the credibilistic fuzzy c-means method is used to appropriately group the data. These algorithms use score vectors of PLS as feature vectors. So, we find total proper different normal operation conditions. To identify operation change modes, a discrimination index is proposed based on the time-series pattern of the membership values of clusters. Using this index, we can monitor all data patterns. The proposed monitoring method is applied to real experimental data from a full-scale power plant process and simulated data from a continuous stirred tank reactor model, and the results are discussed. In particular, the proposed method is found to easily discriminate between intermediate states and faults (or abnormalities) occurring within the process data.

PCB and PAH releases from power stations and waste incineration processes in the UK.

This study focused on emissions of polychlorinated biphenyls (PCB) and polycyclic aromatic hydrocarbons (PAH) from incineration and power generation processes. Increased concern over human exposure to both classes of compounds has meant that environmental regulators need to assess the contribution made by emissions from regulated processes to human exposure. In the first part of an assessment in the UK we reviewed literature data on emissions of PCB, focusing on the dioxin-like PCB assigned toxic equivalency factors by the World Health Organization, and PAH. The literature study was supplemented by a series of plant tests to gather initial real plant data. Literature data were limited and the lack of standard protocols for measurement and reporting of both PCB and PAH meant that few data sets were comparable. Levels of dioxin-like PCB reported in the literature and measured in UK plant tests showed that well-controlled modem combustion plants with comprehensive pollution controls gave low emissions, typically about 5-10% of the toxic equivalent of the emissions of polychlorinated dibenzodioxins and dibenzofurans at the same plants and below the widely used standard of 0.1 ng TEQ/N m3.

Boiler Leak Detection Using a System Identification Technique

Boilers are important processes in chemical and refinery industries: they are normally operated for an extended period of time, leading to (water/steam) tube leaks because of aging and corrosion. To maintain production in normal and safe conditions, detecting the possible boiler leaks in time is crucial. In this paper, a process model is proposed to describe the boiler tube leak problem. On the basis of this model and the boiler characteristics, a least-squares method with a forgetting factor is derived to detect boiler leakage. The analysis of leak estimation properties for the proposed method is given under certain conditions of boiler operation. The applications based on both simulation and real plant data show that the proposed method is capable of detecting boiler leaks effectively and efficiently.

Use of damage monitoring systems for component life optimisation in power plant

This paper presents a number of practical examples where component life monitoring has been implemented on power plants. In particular, case histories are presented for the strategic components of high temperature boiler headers and turbines, subject to the damage mechanisms of creep—due to on load temperatures and pressures, and fatigue—due to start-up and shutdown thermal transient operation. The use of these systems have demonstrated that faster start-ups and more consistent operations can be attained which provide both life extension and enhanced performance in terms of start-up rates. The use of these systems to quantify creep damage is also highlighted. This information from real plant data has a significant impact on remnant life predictions, highlighting hot zones in the boiler and the impact of modified operations to safely enhance plant life. Finally, a cost–benefit example is presented which shows the potential magnitude of benefits that can be attained from optimised operations enabled by component life monitoring.

Dynamic Monitoring Method for Multiscale Fault Detection and Diagnosis in MSPC

A dynamic monitoring method for multiscale fault detection and diagnosis (MFDD) in the wastewater treatment process (WWTP) is proposed. This method is based on dynamic principal component analysis (DPCA) and the D statistic and the monitoring of individual eigenvalues of generic dissimilarity measure (GDM). DPCA method, where the original measurements are lagged, describes a cause and effect of the process. Then the GDM of the DPCA score value, called the D statistic, is used to detect faults and events of the current operating condition. Additionally, monitoring of individual eigenvalues enables the diagnosis of diverse kinds of fault and disturbance sources. The DPCA and the dynamic GDM of the MFDD can be used as a remedy to the problem of analyzing nonstationary processes. The proposed method was applied to fault monitoring and isolation in simulation benchmark data and real plant data. The simulation results clearly show that the method effectively detects faults in a dynamic, multivariate, and multiscale process. Moreover, the proposed approach not only detects faults but also isolates the sources of them to some degree.

On-Line Optimization of a Crude Distillation Unit with Constraints on Product Properties

The optimization of chemical processes has gained immense importance in the present day. This work is aimed at the on-line optimization of an industrial crude distillation unit (CDU). A nonlinear, steady-state CDU model has been developed in-house for this purpose. Model tuning parameters in the form of vaporization efficiencies were incorporated to minimize the discrepancy between the measured and simulated column parameters on-line. The crude feed composition, represented by the true boiling point (TBP) curve, is known to vary with time. A procedure was developed to back-calculate the TBP curve using on-line plant data. Finally, the objective function was formulated to simultaneously maximize the net achievable profit and set the product properties within a user-specified range. The entire scheme was tested using real plant data off-line, but the problem formulation is suitable for supervisory-level on-line optimization without further modifications. It is shown that substantial increases in profitabili...

Improved methodology for generation of axial flux shapes in digital core protection systems

An improved method of axial flux shape (AFS) generation for digital core protection systems of pressurized water reactors is presented in this paper using an artificial neural network (ANN) technique—a feedforward network trained by backpropagation. It generates 20-node axial power shapes based on the information from three ex-core detectors. In developing the method, a total of 7173 axial flux shapes are generated from ROCS code simulation for training and testing of the ANN. The ANN trained 200 data predicts the remaining data with the average root mean square error of about 3%. The developed method is also tested with the real plant data measured during normal operation of Yonggwang Unit 4. The RMS errors in the range of 0.9∼2.1% are about twice as accurate as the cubic spline approximation method currently used in the plant. The developed method would contribute to solve the drawback of the current method as it shows reasonable accuracy over wide range of core conditions.

Teaching desalination

Desalination is a multi-disciplinary engineering science that includes elements of heat, mass, and momentum transfer; thermodynamics; unit operations design; material engineering; energy conversion; water chemistry; corrosion; scale formation and fouling. This paper outlines desalination teaching through four main venues, which includes undergraduate engineering courses, graduate engineering courses, intensive training courses for engineers, and desalination research. The undergraduate and graduate engineering courses are taught as an elective component in the chemical/mechanical engineering curriculum. The contents of the undergraduate course provides the student with an overview of the desalination processes and elements of design, while the graduate course focuses on more detailed design calculations, process synthesis, process control, case studies, and use of computer packages for system design. The intensive training course for engineers includes introductory and advanced levels of instruction. The introductory training course includes historical background, process diagrams of thermal and membrane desalination processes, elements of design, performance parameters, associated processes, practice, and trends. The focus of the advanced training course is on the use of computer packages for system design, evaluation of real plant data, and cost estimation. The research program, which is part of the desalination education process, includes mathematical and experimental investigations that include system modeling, process control, study of fundamental phenomena, energy considerations, process synthesis, and use of new construction materials and chemicals.

Model-Based Estimation of Metal Analysis in the Steel Converter: Linear Versus Nonlinear Approach

The problem of real time metal analysis estimation for the basic oxygen steelmaking process is considered. Nonlinear feedback is applied to a nonlinear proccss model, thus creating a nonlinear observer. It is compared to linear feedback on the same process model and is shown to have superior performance. Using real plant data from the converter at SSAB Oxelösund AB, the observer is shown to provide accurate estimates of the carbon content.

Modeling of industrial naphtha cracking furnaces

The ethylene producing plant is one of the largest plant in chemical industry. The industrial naphtha cracking furnace is the most important unit in the ethylene production. However, the modeling of naphtha cracking furnace is known to be extremely difficult. First, naphtha feedstock consists of a lot of hydrocarbon components that range from C5 to C10 broadly. Numerous thermal cracking reactions with complex mechanism also occur in the reactor tube with transient carbon deposition. Thus a rigorous reactor model which can cope with all these mechanisms is essential for the modeling of cracking furnace. In addition, overall performance of the cracking furnace is mainly determined by the complex heat transfer phenomena between the reator tube and the radiant box that provides the combustion heat to the cracking reactors by fuelfiring. Therefore, a rigorous model that combines the radiant box and the reactor is very important for detail analysis of the cracking furnace. In this study, we develope a rigorous cracking furnace model which combines the radiant box and the reactor. The model is not only as flexible to any reactor configuration in radiant box but also as rigorous to calculate the detail profiles of product compositions and temperature along the reactor length. The radiant box is modeled by single gas zone method. The carbon deposition model uses the relative rates of coke production for several precursors selected by isotope experiments. The reactor is modeled as a general plug flow reactor. The developed model can be tuned easily with plant operating data to incorporate the detail characteristics of process equipment and coke deposition. The model is validated using real plant data and open literatures for various reactor configurations and for both naphtha and ethane feeds and used to analyze the process through several challenging case studies.

Issues regarding artificial neural network modeling for reactors and fermenters

In recent years researchers in many areas have used artificial neural networks (ANNs) to model a variety of physical relationships. While in many cases this selection appears sound and reasonable, one must remember than ANN modeling is an empirical modeling technique (based on data) and is subject to the limitations of such techniques. Poor prediction occurs when the training data set does not contain adequate “information” to model a dynamic process. Using data from a simulated continuous-stirred tank reactor, this paper illustrates four scenarios: (1) steady state, (2) large process time constant, (3) infrequent sampling, and (4) variable sampling rate. The first scenario is typical of simulation studies while the other three incorporate attributes found in real plant data. For the cases in which ANNs predicted well, linear regression (LR), one of the oldest empirical modeling techniques, predicted equally well, and when LR failed to accurately model/predict the data, ANNs predicted poorly. Since real plant data would resemble a combination of situations (2), (3), and (4), it is important to understand that empirical models are not necessarily appropriate for predictively modeling dynamic processes in practice.

Mass estimation using a loose parts monitoring system

This paper is concerned with estimating the mass of a loose part in the steam generator of nuclear power plant. Although there is the basic principle known as “Hertz Theory” for estimating mass and energy of a spherical part impacted on an infinite flat plate, the theory is not directly applicable because real plants do not comply with the underlying ideal assumptions. In this work, a practical method is developed based on the basic theory and considering amplitude and energy attenuation effects. Actually, the impact waves propagating along the plate to the sensor locations become modified in shape and frequency spectrum from the original waveform due to the plate and surrounding conditions, distance attenuation and damping loss. To show the validity of the present mass estimation algorithm, it has been applied to the mock-up impact test data and also to the real plant data. The results show better performance comparing to the conventional Hertz schemes.

Development of High Precision Plant Simulator for Pressurized Water Reactor Plants using Distributed Architecture

As advanced monitoring and controlling systems have been developed, accuracy of real-time simulators must be improved and simulation limits must be extended. Therefore the authors have developed a distributed simulation system to achieve high processing performance using low cost hardware. Moreover, the authors have developed a thermal-hydraulic computer code, using drift-flux non-equilibrium model. This code can realize a high precision two-phase flow analysis, to be considered to have the same prediction capability as two-fluid models, while achieving high speed and stability for real-time simulators. The distributed plant simulator for PWR plants was realized as a result. The distributed simulator consists of multi-processors connected to each other by an optical fiber network. Controlling software for synchronized scheduling and data transfer was also developed. In terms of the code development, the numerical calculation method and implemented correlation chosen for this code are described. The simulation results during both code development and validation testing of the four loop PWR simulator are compared with experiment data and real plant data; the agreement was satisfactory for a plant simulator. The simulation speed was also satisfactory being twice as fast as real-time.

Intelligent on-line system identification using a blackboard system

Adaptive control systems have been developed and used for a number of decades. However, there are still some problems with their operation, which limit wider industrial adoption. This paper addresses one aspect of adaptive control, namely the on-line system identification of the time-varying process being controlled. The approach adopted is to use a blackboard system to identify the time-varying process model. The blackboard system contains knowledge sources with, algorithmic, fuzzy-logic and evolutionary reasoning. The paper describes the design of a computer simulation of this approach, concentrating on the fuzzy reasoning used to validate the multiple models, and the evolutionary techniques used to reject poor models and introduce better models. The computer simulation is then evaluated, firstly using simulated plant data and then using real plant data from an experimental hot air dryer. It is concluded that the technique is practical, and worthy of further testing on a pilot plant.

The problem of too much power in detecting biases for a real chemical process

Statistical hypothesis testing in a strategic fashion is often used to locate biased process variables and nodes (interconnecting points) with material leaks. Power, which increases with the number of measurements used in a hypothesis test, represents the ability of the test to correctly conclude the existence of bias or a leak. As computer technology continues to improve, so does the ability to store large quantities of data. Measurement bias occurs from miscalibrated instruments and equipment failures. To some degree, most instruments will have some amount of bias at any given time. Thus, with huge data sets, statistical hypothesis testing strategies will flag nearly all instruments as being biased, which is not a practical result when one is only interested in a bias above a certain threshold (i.e., control of power). This work demonstrates the dilemma of too much power and stresses the control of power by using power functions and a controlled number of measurements. This study was conducted using real plant data provided by the Shell Development Company.

Rapid estimation of floatability components in industrial flotation plants

In the analysis of industrial flotation plant data, it is advantageous to build a reliable computer based flotation model to be used in optimisation and design. A number of models have been proposed by many authors with each model having value depending on the complexity of the situation for which it was developed. Several of these models assume that a floatability distribution of a mineral entering a flotation circuit can be described by a number of floatability classes or components with each of these components (and its flotation rate constant) being conserved around the circuit. A non-linear optimisation procedure for regressing floatability parameters (number of floatability components, and the flotation rate constant of each) from laboratory batch flotation tests conducted on streams around the circuit has recently been devised [1]. This paper outlines a technique to linearise the above regression procedure which improves the ease of solution and the level of confidence in the floatability parameter determination. The technique also offers a strategy for determining the “optimum” number of floatability components. Dummy and real plant data were used to test the validity of the new linear modelling technique. The dummy data were generated using a set flowsheet and a feed stream having two components (with set first order rate constants) to simulate the plant. The simulated data obtained were then fitted with the general linear model technique. The results of this operation show that the linear modelling procedure can obtain the original model parameters very well. The model was then tested on real plant data from three typical flotation plants. Again, there is a very good correlation between stream experimental and model calculated recoveries based on the feed to the circuit. The confidence limits of the model parameters generated using this technique are reported. The ability to estimate parameter confidence intervals indicates where additional data (or an independent method) are required for more accurate parameter determination.

MULTIVARIABLE CONSTRAINED ADAPTIVE GPC: THEORY AND EXPERIMENTAL EVALUATION

This paper deals with on-line identification and constrained long-range predictive control of multivariable systems. It extends a recently proposed augmented upper diagonal factorization identification (AUDI) algorithm to identify input-output models of multivariable systems with distinct time delays. The multi-input, multi-output (MIMO AUDI) algorithm can simultaneously identify the process model order and process parameters. The MIMO AUDI algorithm is implemented by decomposing a MIMO system into as many multi-input, single-output (MISO) subsystems as the number of outputs and then identifying each MISO subsystem separately. The performance of the new MIMO AUDI algorithm is demonstrated by application to input-output data from a real process. The extension of this algorithm by incorporating a variable forgetting factor with a lower bound in its value is implemented on real plant data to demonstrate 'alertness' of the estimator. This paper evaluates the performance of the MIMO adaptive generalized predictive control algorithm with and without constraints by experimental application on a computer-interfaced, pilot-scale process. The MIMO adaptive GPC is shown to have good regulatory plus servo-tracking properties.

A VLSI Image Processing Architecture Dedicated to Real-Time Quality Control Analysis in an Industrial Plant

In this paper, we present a VLSI architecture for real-time image processing in quality control industrial applications: automation of the visual inspection phase of mechanical parts treated by the Fluorescent Magnetic Particle Inspection method for structural-defect detection. The VLSI architecture implements a highly constrained neural network tailored for this specific application: the multi-layer perceptron with strictly local connections. The learning of the weights is performed off line by using the adaptive simulated-annealing algorithm. The neural network has been trained on real plant data: recognition results of the training and classification tasks compare favorably with those obtained by expert human operators.The VLSI architecture receives as input the image (taken on-line on the plant) of a mechanical part and it will find out if at least one structural surface defect is present. The VLSI architecture was optimized, through a set of transformations on the high-level VHDL specifications of the neural network algorithm, to reach real-time operating conditions. Following the proposed approach and the designed architecture, we designed and successfully tested a custom VLSI chip for the real-time implementation of the recognition task.

Distillate purity from MSF: The theoretical design and a real case behaviour

The purity of distillate water coming from MSF desalination plants is of its most importance when the water is used for industrial services. The use of a correct distillate purity design can reduce the operating costs and optimizes the investiment cost. This work aims at describing the different theories which regulate the salt entrainment and the distillate purification, and at supporting them with a real case plant experimental data. Moreover a theoretical equation for demister efficiency is proposed whose behaviour is correlated to the real plant data in order to set parameters which can be used for an optimal stage and demister design.