Search

• First integrated optimization of crude oil procurement and blending. • Proposing a hybrid time-event representation to address different time horizons. • Considers the stability of blended crude oil properties in the optimization process. • Verification of the authenticity and validity of the model on the basis of real cases. • Optimization results in effective reduction of crude oil procurement costs. Crude oil procurement and blending are key processes in refinery production. When solving the integrated optimization problem of simple oil procurement and blending, the following issues are mainly considered: how to establish an integrated optimization model of crude oil procurement and blending process under different time scales to minimize the cost of procurement under the premise of ensuring the stability of the properties of blended crude oil. The paper proposes an integrated optimization model, takes cost minimization as the model's objective function, and provides stability through the yield and property constraints of blended crude oil. Then, the paper describes the procurement process by an event-based representation and the blending process by a continuous event representation and proposes a hybrid event-time-based representation to describe the model. Finally, this paper verifies the model's effectiveness in solving real production problems through case simulations. [ABSTRACT FROM AUTHOR]

Oscillations in control loops are one of the most prevalent problems in industrial processes. Due to their adverse effect on the overall process performance, finding how oscillations propagate through the process units is of major importance. This paper presents a method for integrating process causality and topology which ultimately enables to determine the propagation path of oscillations in control loops. The integration is performed using a dedicated search algorithm which validates the quantitative results of the data-driven causality using the qualitative information on plant connectivity. The outcome is an enhanced causal model which reveals the propagation path. The analysis is demonstrated on a case study of an industrial paperboard machine with multiple oscillations in its drying section due to valve stiction. [ABSTRACT FROM AUTHOR]

While machine learning (ML) has made significant contributions to the biopharmaceutical field, its applications are still in the early stages in terms of providing direct support for quality-by-design based development and manufacturing of biologics, hindering the enormous potential for bioprocesses automation from their development to manufacturing. However, the adoption of ML-based models instead of conventional multivariate data analysis methods is significantly increasing due to the accumulation of large-scale production data. This trend is primarily driven by the real-time monitoring of process variables and quality attributes of biopharmaceutical products through the implementation of advanced process analytical technologies. Given the complexity and multidimensionality of a bioproduct design, bioprocess development, and product manufacturing data, ML-based approaches are increasingly being employed to achieve accurate, flexible, and high-performing predictive models to address the problems of analytics, monitoring, and control within the biopharma field. This paper aims to provide a comprehensive review of the current applications of ML solutions in the design, monitoring, control, and optimisation of upstream, downstream, and product formulation processes of monoclonal antibodies. Finally, this paper thoroughly discusses the main challenges related to the bioprocesses themselves, process data, and the use of machine learning models in monoclonal antibody process development and manufacturing. Moreover, it offers further insights into the adoption of innovative machine learning methods and novel trends in the development of new digital biopharma solutions. • Provide a review of applications of ML in antibody design and manufacturing. • Identifying challenges related to bioprocesses and process data. • Offering a valuable reference with a thorough discussion and public datasets. • Proposing machine learning (ML) research directions in antibody manufacturing. • Identifying advanced ML solutions for digital twins within biopharma 4.0. [ABSTRACT FROM AUTHOR]

• Developed extension of the acceptance-rejection method for identification of the design space (DS) in a two-stage batch reactor system. • Studied the time dynamics and the effect of critical quality attributes (CQAs) on critical process parameters (CPPs) between different stages. • Studied different settings of imposed CQAs and their connections with tight operations versus balanced operations. • Employing the potential second stage DS volume as a quality metric enables adjusting DS sizes and transition between tight and balanced operations. The design space (DS) is defined as the combination of materials and process conditions that guarantees the assurance of quality. This principle ensures that as long as a process operates within DS, it consistently produces a product that meets specifications. It was originally developed for a single unit system. Many industrial processes frequently involve multiple unit operations. Assessing the interaction of Critical Process Parameters (CPPs) with Critical Quality Attributes (CQAs) across stages enables informed decision-making and the capacity to balance different requirements. Analysis and visualization of the complex, multi-dimensional DS is a challenging task. This paper presents a framework for identification such DSs, considering both joint and decoupled strategies using a detailed analysis of a two-stage batch reactor case study. We assess and discuss the practicality and relevance of these methods. [ABSTRACT FROM AUTHOR]

• Large-scale combined scheduling and energy procurement optimization problems are solved. • Energy-cost optimization and energy-aware production scheduling are functionally separated. • Different variants of Mean Value Cross Decomposition are investigated that yield near-optimal or optimal solutions. • Existing solutions for the scheduling and for the energy procurement problem that take into account problem-specific aspects can be integrated. • Energy-intense industries as e.g. steel and pulp and paper can reduce their energy cost while meeting the production demands. Integrated optimization of the procurement cost of electric energy with production planning is increasingly considered in various industries. The traditional approach in industry is production driven, i.e. the production is scheduled first, followed by the energy supply optimization to find the best available energy portfolio, which is usually sub-optimal. The combined scheduling and energy procurement optimization can be formulated as an integrated monolithic optimization model, resulting in intractable problems, even if solutions to the two isolated problems are available. We propose to use Mean Value Cross Decomposition for solving the combined problem by iterating between energy-aware production scheduling and energy-cost optimization, possibly building on existing solutions. We apply the approach to a pulping process and a steel production process. MILP-based models are employed for the two scheduling problems and for the energy cost optimization a Minimum-Cost Flow Network model is used, resulting in good quality solutions within reasonable computation times. [ABSTRACT FROM AUTHOR]

• Pharmaceutical process design involves much uncertainty in the data on which it is based. • The paper presents a quantitative approach to risk management for design using a stochastic approach which identifies key data required and allows for its systematic incorporation. • Expected performance was improved by considering the propagation of uncertainty over the whole process. • In this case study it was found that below a certain value in the feed purity 'perfect' knowledge had a very significant impact. Process plants for the manufacture of pharmaceutical products often need to be designed and built quickly to make the most of available patent life which necessitates using uncertain or unavailable data. It is common that pilot plant equipment and data are available and new data can be generated if they are important. We present a model based approach to risk analysis to aid design for pharmaceutical processes which combines systematic modelling procedures with Hammersley sampling based uncertainty analysis and sensitivity analysis used to quantify predicted performance uncertainty and to identify key uncertainty contributions. The main contribution of the paper is the demonstration of the methodology on an industrial case study where the process flowsheet was fixed and some pilot data was available. Expected performance was improved by considering the propagation of uncertainty over the whole process. The case study results indicate the importance of considering uncertainty systematically and quantitatively. The methodology showed the opportunity to improve process performance potential through considering uncertainty systematically. [ABSTRACT FROM AUTHOR]

Big data and time series are typical features of modern industrial process data. Effective time series modeling methods are required for ensuring the normal and stable operation of industrial processes. However, the existing time series prediction methods usually cannot take important historical data features long ago into consideration. In this paper, we proposed a new attention-mechanism based dynamic-inner partial least squares long short-term memory (ADiPLS-LSTM) method to predict the time series data. First, the DiPLS method is applied to extract the dynamic features in the selected data group, and the attention-mechanism is used then to define the importance of the responded features. The product of the extracted dynamic features and the attention-mechanism results is regarded as the input of LSTM to predict the future time information. The proposed method uses both the recent data information and important data information in a long run, which helps to obtain more accurate prediction results. The effectiveness of the ADiPLS-LSTM method is verified by the data of a 660MW coal-fired boiler. Compared with BPNN, LSTM, and DiPLS-LSTM, the proposed method also advances in time series data prediction and interpretability. • The receive field of LSTM is usually very limited, and the interpretability is weak • DiPLS can extract dynamic features between input and output • The existing prediction methods cannot take important historical data features. • A new novel ADiPLS-LSTM method is proposed in this paper. • Application results confirm the effectiveness and reliability of the proposed method. [ABSTRACT FROM AUTHOR]

• Optimized a novel loss function ESE to quantify shape differences in chemical process data. • Proposed EPMITS method to learn both trend and shape features by two training steps. • Evaluated performance of EPMITS method under a real fluid catalytic cracking dataset. • High prediction accuracy and low training time cost with potential industrial applications. With the transformation of industrial production digitization and automation, process monitoring has been an indispensable technical method to realize the safe and efficient production of chemical process. Accurate prediction of process variables in chemical process can indicate the possible system change to reduce the probability of abnormal conditions. Current popular deep learning prediction methods trained with MSE or its variants may exhibit limitations in extracting shape features of chemical process data. In this paper, we proposed an efficient prediction method incorporating trends and shapes features (EPMITS) for chemical process variables. Specifically, we introduced a novel differentiable loss function Efficient Shape Error (ESE) to quantify shape differences between two time series of equal length in chemical process data. Then we trained deep learning models with MSE and ESE as loss function by two steps in training stage, to effectively acquire both trend and shape features of chemical process data. The proposed method was evaluated by the Tennessee Eastman process datasets and a real fluid catalytic cracking dataset from a petrochemical company. The results indicate that EPMITS models exhibit high prediction accuracy and short model training time across various time scales. These findings demonstrate the considerable feasibility and significant potential of EPMITS for future fault prognosis applications. [Display omitted] [ABSTRACT FROM AUTHOR]

• A data enhancement method based on generative adversarial network is proposed. • The method combines the advantages of data augmentation and feature selection. • Case studies on a simulated case and two real industrial processes are provided. • Comparisons to state-of-the-arts show effectiveness of the proposed method. Soft sensor plays an important role in improving product quality; however, practical applications may often face with the problem of small sample size, which is challenging for developing data-driven models in terms of feature selection and good generalization. This paper proposes a data enhancement approach for small sample size data-driven problems based on generative adversarial networks integrated with maximum relevance minimum redundancy (MRMR). First, sample expansion is performed on the initial data by using a generative adversarial network. Second, irrelevant variables are eliminated by the MRMR and optimal features are obtained. Finally, neural networks-based soft sensor modeling is performed using the augmented dataset and the selected features. The proposed method is tested on a simulated penicillin case, an actual penicillin production case and an actual erythromycin production case. Experimental results show that the proposed method outperforms state-of-the-art existing methods, which verify the effectiveness and superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Bayesian optimization (BO) has emerged as a useful paradigm for automatic calibration (aka auto-tuning) of advanced optimization- and learning-based controllers whose closed-loop performance depends on the choice of several tuning parameters in highly nonlinear and nonconvex ways. However, BO approaches to controller auto-tuning commonly rely on the assumption that system dynamics remain constant, which does not hold for systems with time-varying dynamics, for example, due to gradual aging or persistent environmental drifts. This challenge can be further compounded when gradual and persistent system drifts occur over a series of process runs. Existing time-varying BO (TVBO) approaches with spatio-temporal kernels fall short of effectively handling an integer run index, which is imperative for capturing run-to-run changes in the system behavior. To this end, this paper presents a run-indexed TVBO (RI-TVBO) approach that can systematically account for run-to-run process drifts as the system is queried over sequential process runs. The proposed approach relies on adapting the non-stationary Wiener process kernel to accommodate an integer run index, instead of time. This is done via positional encoding that incorporates the integer run index and, thus, enables describing run-to-run variations in system dynamics. The positional embedding vector associated with each run index is then mapped onto a scalar value to leverage the relationships between different process runs within the probabilistic surrogate model of the objective function in RI-TVBO. The performance of RI-TVBO is evaluated for auto-tuning of an offset-free model predictive controller for a low-temperature plasma-assisted process for thin film deposition. Simulation results demonstrate the superior performance of RI-TVBO over standard BO and TVBO under different scenarios of run-to-run process drifts encountered in plasma-assisted deposition processes in semiconductor manufacturing. • A run-indexed time-varying Bayesian optimization method is presented that can systematically handle run-to-run system drifts and variations. • The performance of the proposed method is demonstrated for auto-tuning of an offset-free MPC controller. • Simulation results demonstrate the effectiveness of the proposed method in handling run-to-run process drifts that commonly occur in semiconductor manufacturing. [ABSTRACT FROM AUTHOR]

Abstract: Optimization of the steady state economic efficiency of an industrial process is a specific task because the decision variables of the optimization (setpoints of the control system) affect the process through the control strategy. Thus, the effects of saturation of a control system must be taken into account when the gradient of the objective function is estimated and the necessary optimality conditions are checked. In particular, because the optimality conditions cannot be checked directly in the presence of active constraints on the manipulated variables, approximations of the steady state values of the manipulated variables as functions of the setpoints (static plant model) are needed in order to be able to evaluate the optimality conditions. In this paper an iterative method for optimization of the plant profit rate is proposed avoiding the control saturation and is applied to the Pulp Mill benchmark model optimization. Three different static models describing the steady state values of the manipulated variables are constructed and used in the optimization. The results of the optimization are presented and compared against the straightforward single-step optimization of the plant economic efficiency. [ABSTRACT FROM AUTHOR]

This paper presents a multi-week mixed integer linear programming (MILP) scheduling model for an ice cream processing facility. The ice cream processing is a typical complex food manufacturing process and a simplified version of this processing has been adapted to investigate scheduling problems in the literature. Most of these models only considered the production scheduling for a week. In this paper, multi-week production scheduling is considered. The problem has been implemented as an MILP model. The model has been tested on a set of cases from the literature, and its results were compared to the results of problems solved using hybrid MILP-heuristics methods in the literature. The inclusion of clean-up session, weekend break and semi-processed product from previous week were also assessed with two additional sets of experiments. The experiments result show that the proposed MILP is able to handle multi-week scheduling efficiently and effectively within a reasonable time limit. [ABSTRACT FROM AUTHOR]

To cope with a cost-effective manufacturing approach driven by more than Moore’s law era, plasma etching which is one of the major processes in semiconductor manufacturing has developed plasma sensors and their applications. Among the plasma sensors, optical emission spectroscopy (OES) has been widely utilized and its high dimensionality has required multivariate analysis (MVA) techniques such as principal component analysis (PCA). PCA, however, might devaluate physical meaning of target process during its statistical calculation. In addition, inherent noise from charge coupled devices (CCD) array in OES might deteriorate PCA model performance. Therefore, it is desirable to pre-select physically important variables and to filter out noisy signals before modeling OES based plasma data. For these purposes, this paper introduces a peak wavelength selection algorithm for selecting physically meaningful wavelength in plasma and discrete wavelet transform (DWT) for filtering out noisy signals from a CCD array. The effectiveness of the PCA model introduced in this paper is verified by comparing fault detection capabilities of conventional PCA model under the various source power or pressure faulty situations in a capacitively coupled plasma etcher. Even though the conventional PCA model fails to detect all of the faulty situations under the tests, the PCA model introduced in this paper successively detect even extremely small variation such as 0.67% of source power fault. The results introduced in this paper is expected to contribute to OES based plasma monitoring capability in plasma etching for more than Moore’s law era. [ABSTRACT FROM AUTHOR]

This paper introduces an innovative method for batch industrial process monitoring and fault detection. By combining time-slice dynamic prediction with an enhanced adaptive multiway principal component analysis (PCA) technique, it aims to balance the trade-off between high fault detection rates and low false alarm rates in monitoring batch industrial data. The method extracts dynamic features from 3D batch data time slices and utilizes a serial PCA framework with variable control limits for real-time monitoring. Comparative evaluations against traditional adaptive multiway PCA and dynamic time-slice canonical correlation analysis approaches demonstrate its superior performance. A case study using a novel cold rolling mill simulation model underscores its effectiveness for batch industrial process monitoring and fault detection. • Novel approach for dynamic extraction from batch industrial process data. • Time-slice monitoring and detection scheme based on serial PCA-KPCA. • Method validated on a novel cold rolling mill model. [ABSTRACT FROM AUTHOR]

Increased globalization, deregulation of energy markets, and environmental constraints, together with associated uncertainty, have created a highly dynamic and uncertain process manufacturing environment. Responding effectively to this increased variation and uncertainty is critical for a company to remain competitive. In this paper, we consider the plant infrastructure in relation to the energy and global market infrastructures. We describe changes to the plant infrastructure system in order to function more effectively in the current manufacturing environment, as well as key research advances that are aligned to addressing challenges faced by present day plant operation. • Discusses relationship between plant, energy, and global market infrastructures. • Identifies challenges associated with the current plant operating environment. • Discusses industry response to identified challenges. • Overview of PSE tools and paradigms aligned with addressing these challenges. • PSE approaches illustrated in the context of industrial air separation plants. [ABSTRACT FROM AUTHOR]

As manufacturing processes transition towards digitalization, data-driven process control is emerging as a key area of interest in future artificial intelligence technology. A crucial aspect in implementing data-driven process control is "What should we learn from the data?". In general, the data-driven control method can be categorized into two main approaches: Learning the model and learning the value. To assist in selecting the more suitable approach, this paper applies six different control methods, with three falling under each approach, to three distinct manufacturing process systems. The simulation results indicate that the model-learning approaches display higher data efficiency and exhibit lower variance in total cost. These methods prove to be particularly advantageous for addressing the regulation problems. Conversely, value-learning approaches show competitive potential in closed-loop identification and in managing economic cost problems. The remaining challenges associated with each technique are discussed, along with practical considerations for their implementation. • What should we learn, model or controller, with the collected data? • Quantitatively compare the model-based and model-free control approaches. • Model-based methods were superior in tackling regulation problems. • Model-free methods show competitive performance in optimizing economic cost. [ABSTRACT FROM AUTHOR]

The electrical energy (EE) price plays a significant role in the economic assessment of industrial processes. PSE/CAPE applications are often based on economic evaluations/optimizations where EE price and its possible dynamic evolution are important input data. Planning, scheduling, on-line optimization (dynamic) conceptual design, and feasibility studies are some of the applications where short-, medium-, and long-term predictions of EE quotations are involved. The paper discusses the main issues that contribute to EE quotations such as geographical, meteorological, seasonal, political, social, and financial terms. EE prices show a significant dependency on crude oil quotations with a time-delay of about a quarter/season. An econometric model comprising both linear and periodic components with an implicit stochastic term delivered by the reference component is first commented, then identified, and finally validated respect to real EE quotations. The paper provides also a discussion on how to forecast EE prices ranging from short- to long-term horizons. [ABSTRACT FROM AUTHOR]

• Summarize sustainability challenges and future trends of energy systems. • Analyze various development options and their impact on the power grid. • Propose multi-disciplinary research steps and challenges that must be addressed. The increased awareness of the climate change has triggered numerous actions and world-wide policies (e.g. the Paris agreement), with strong incentives to stop using fossil fuel-based energy forms in order to achieve a climate neutral economy (OECD, 2019). To realize the political targets, there are numerous challenging technical hurdles to overcome, which call for continued research. Earlier isolated systems need to come together and jointly collaborate to enable many changes in processes, for instance, seeking opportunities for electrification of energy intensive processes (Vogl et al., 2018). In this quest power systems play a crucial role through the increasing electricity transmission volumes, and they are already in the process of fundamental shift towards climate neutrality. The scale and urgency of this transformation is enormous and will affect most industries due to the central role of energy and electricity. In this paper, we discuss the main ongoing developments and future trends that also imply a closer coordination between power systems and various process systems. [ABSTRACT FROM AUTHOR]

This paper examines the prediction of disturbances based on their past measurements using k -nearest neighbours. The aim is to provide a prediction of a measured disturbance to a controller, in order to improve the feed-forward action. This prediction method works in an unsupervised way, it is robust against changes of the characteristics of the disturbance, and its functioning is simple and transparent. The method is tested on data from industrial process plants and compared with predictions from an autoregressive model. A qualitative as well as a quantitative method for analysing the predictability of the time series is provided. As an example, the method is implemented in an MPC framework to control a simple benchmark model. [ABSTRACT FROM AUTHOR]

In the process industry, various types of degradation occur in processing plants, resulting in significant economic losses. Modeling of degradation is important because it provides quantitative insights for consideration of degradation impacts in the operations of process manufacturing. This paper studies batch processes that show a periodic pattern for the evolution of degradation. A new data structure, the campaign, is applied for data-driven modeling of the periodic batch-to-batch evolution of degradation using a new multiway partial least squares approach, and it is further employed to predict the evolution of degradation in a series of batch runs. The proposed approach is illustrated and applied in a comprehensive industrial case study. The example illustrates the efficacy of the proposed model and presents a fair potential for applications of degradation prediction. [ABSTRACT FROM AUTHOR]

• Multi-objective planning under uncertainty using a continuous-time heuristic model. • Monte Carlo simulation is integrated with a variable-length multi-objective GA. • Monte Carlo simulation is accelerated with a Graphics Processing Unit (GPU). • The proposed approach is validated on an industrially-relevant case study. • The benefits of the approach are demonstrated by comparison with a deterministic GA. This paper presents a flexible genetic algorithm optimisation approach for multi-objective biopharmaceutical planning problems under uncertainty. The optimisation approach combines a continuous-time heuristic model of a biopharmaceutical manufacturing process, a variable-length multi-objective genetic algorithm, and Graphics Processing Unit (GPU)-accelerated Monte Carlo simulation. The proposed approach accounts for constraints and features such as rolling product sequence-dependent changeovers, multiple intermediate demand due dates, product QC/QA release times, and pressure to meet uncertain product demand on time. An industrially-relevant case study is used to illustrate the functionality of the approach. The case study focused on optimisation of conflicting objectives, production throughput, and product inventory levels, for a multi-product biopharmaceutical facility over a 3-year period with uncertain product demand. The advantages of the multi-objective GA with the embedded Monte Carlo simulation were demonstrated by comparison with a deterministic GA tested with Monte Carlo simulation post-optimisation. [ABSTRACT FROM AUTHOR]

• A new deterministic model for WDN optimization is proposed considering unknown flow directions. • The minimum WDN cost model considers a set of available commercial diameters. • No hydraulic simulator is used for calculating pressures and velocities. • Disjunctive programming is used in the model formulation. • Global optimum solutions were obtained. Water Distribution Networks (WDN) are present in a large number of industrial processes and urban centers. Reservoirs, pipes, nodes, loops, and pumps compose WDN and their design can be formulated as an optimization problem. The main objective is the minimization of the network cost, which depends on the pipe diameters and flow directions known a priori. However, in the design of new WDN in real industrial problems, flow directions are unknown. In the present paper, a disjunctive Mixed Integer NonLinear Programming (MINLP) model is proposed for the synthesis of WDN considering unknown flow directions. Two case studies are employed to test the model and global optimization techniques are used in its solution. Results show that the global optima WDN cost with the correct flow directions is obtained for the studied cases without the necessity of using additional software to calculate pressure drops and velocities in the pipes. [ABSTRACT FROM AUTHOR]

Highlights • Upstream plantwide modelling and optimisation of the continuous pharmaceutical manufacturing (CPM) of atropine. • Continuous liquid–liquid extraction (CPM-LLE) design considers ternary phase composition modelling (UNIFAC) and estimated solute partitioning between product LLE phases. • Technoeconomic optimisation for comparative evaluation of separation solvents for CPM-LLE under varying design assumptions (plant capacity, solvent recovery). • Toluene emerges as the most economically viable CPM-LLE solvent choice compared to the batch LLE designs, followed by diethyl ether and n -butyl acetate. • Toluene usage allows acceptable E -factors for pharmaceutical processes. Abstract Continuous Pharmaceutical Manufacturing (CPM) can revolutionise industrial efficiency via potential operational and economic benefits over currently dominant batch methods. Process modelling and optimisation are valuable towards rapid design space evaluation and elucidation of optimal process design configurations, without expensive and time-consuming experimental campaigns. This paper pursues total cost minimisation via nonlinear optimisation of different separation design options for atropine, a product of great societal importance. The study considers a demonstrated continuous flow synthesis, presents reaction kinetic parameter estimation towards reactor design, and illustrates a comparative analysis of the subsequent batch vs. continuous liquid–liquid extraction (LLE) for product purification, using published partition coefficient data and UNIFAC-modelled ternary liquid–liquid equilibria. Original optimisation results show that toluene is the best continuous LLE solvent, attaining the lowest total costs at both plant capacities considered and the greatest total cost savings with respect to the batch LLE design for varying solvent recovery, at acceptable material efficiencies. [ABSTRACT FROM AUTHOR]

• Concept of a software tool optimized for hazard identification is demonstrated. • Semi-automatic HAZOP analysis was performed for an ammonia synthesis plant. • Steady state multiplicity was successfully identified in Aspen HYSYS environment. • An appropriate prediction of complex fault propagation paths was achieved. Process safety is of major importance in chemical industry. Numerous activities have targeted the modification of conventional risk assessment strategies by computer aided approach. In this paper, a software tool for Hazard and Operability (HAZOP) study based on process simulation is presented. Individual components of the proposed software tool are described and the principal methodology of their function is explained. As the simulation engine, commercial process simulator Aspen HYSYS was employed. Proposed tool was applied to a case study of an ammonia synthesis plant based on an existing plant. Hazardous events and operability problems in the syngas purification unit and ammonia synthesis loop have been detected and reported. The steady state multiplicity phenomenon in the ammonia synthesis loop has also been successfully identified. Based on simulation data evaluation performed in the semi-automatic manner by the proposed tool, a HAZOP-like report containing HAZOP deviations and their causes and consequences was generated. [ABSTRACT FROM AUTHOR]

Highlights • A novel data-driven model predictive controller for uni-axial rotomolding is proposed. • Subspace identification algorithm is used to identify the process model. • Proposed approach is implemented on an experimental setup. • Results demonstrate improved quality, variability rejection and ability to achieve product. Abstract This paper addresses the problem of achieving tight product consistency and enabling automated process changes to deliver user-selected criterion based product in a complex industrial batch process such as uni-axial rotational molding. To this end, a data driven state-space model is first identified. For a given trajectory of input moves (heater and compressed air profiles), this dynamic model is able to predict the evolution of the measured variable (internal product temperature). The dynamic model is augmented with a quality model, which relates its own terminal predictions to a selection of key quality variables (sinkhole area, ultrasonic spectra amplitude, impact test metric and viscosity). The dynamic and quality models are in turn utilized within a model predictive control (MPC) framework that enables specifying product quality requirements explicitly. Experimental results demonstrate the ability of the MPC not only in achieving tight quality control but also providing on-spec product for a new specification. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, the focus will be on approximating nonlinear large scale mathematical model of process systems using full order block-structured model. Further, the objective is to achieve a reduced order model for the nonlinear large model with reduced computational complexity, while at the same time being the good approximation of nonlinear model. The modeling approach used for this purpose is block structure models. Input–output Hammerstein structure referred to the classical Hammerstein model has been extended to new Hammerstein structure making use of states and inputs, hence called as input-state (IS) Hammerstein structure. In this paper it is shown that expansion of Taylor series leads to IS-Hammerstein structure. The accuracy of the approximation is improved by including higher order (second order) approximation. The input-state Hammerstein structure provides opportunities for model reduction in context of reducing the computational load by order reduction of states and Jacobians. IS-full order Hammerstein model has been implemented on a case study from the process industry namely the high purity distillation column. Within the operational domain of a process, the IS-Hammerstein structure provides a reduced order mode that can be used for online application purposes (i.e., optimization, model predictive control, etc.). [Copyright &y& Elsevier]

Abstract: This paper considers multivariate statistical monitoring of batch manufacturing processes. It is known that conventional monitoring approaches, e.g. principal component analysis (PCA), are not applicable when the normal operating conditions of the process cannot be sufficiently represented by a multivariate Gaussian distribution. To address this issue, Gaussian mixture model (GMM) has been proposed to estimate the probability density function (pdf) of the process nominal data, with improved monitoring results having been reported for continuous processes. This paper extends the application of GMM to on-line monitoring of batch processes. Furthermore, a method of contribution analysis is presented to identify the variables that are responsible for the onset of process fault. The proposed method is demonstrated through its application to a batch semiconductor etch process. [Copyright &y& Elsevier]

• This paper presents a retrospective comparison of seven representative dynamic latent variable (DLV) models for dynamic process monitoring. • This paper summarizes the essential contributions and problems of representative DLV methods and provides a significant reference for further development on DLV models. • This paper presents guidance on the selection of DLV models in practical application by evaluating the monitoring performance of the discussed approaches through a benchmark process. Process dynamic behaviors resulting from closed-loop control and the inherence of processes are ubiquitous in industrial processes and bring a considerable challenge for process monitoring. Many methods have been developed for dynamic process monitoring, of which the dynamic latent variables (DLV) model is one of the most practical and promising branches. This paper provides a timely retrospective study of typical methods to fill the void in the systematic analysis of DLV methods for dynamic process monitoring. First, several classical DLV methods are briefly reviewed from three aspects, including original ideas, the determination of parameters, and offline statistics design. Second, a discussion on the relationships of the discussed methods has been established to make a clear understanding of process dynamics explained by each method. Third, five cases of a three-phase flow process are provided to illustrate the effectiveness of the methods from the application viewpoint. Finally, future research directions on dynamic process monitoring have also been provided. The primary objective of this paper is to summarize the prevalent DLV methods for dynamic process monitoring and thus highlight a valuable reference for further improvement on DLV models and the selection of algorithms in practical applications. [ABSTRACT FROM AUTHOR]

Abstract: This paper addresses the design and analysis of multisite integration and coordination strategies within a network of petroleum refineries using different crude combination alternatives. In addition, production capacity expansion requirements are also accounted for. The main feature of the paper is the development of a methodology for simultaneous analysis of process network integration alternatives in a multisite refining system through a mixed-integer linear program (MILP) with the overall objective of minimizing total annualized cost. The State Equipment Network (SEN) representation was used for modeling the network as it provides a consistent modeling strategy and proper handling of units that operate under different operating modes, which is common in the refining industry. The integrated network design specifically addresses intermediate material transfer between processing units at each site. The performance of the proposed model was tested on several industrial-scale examples to illustrate the economic potential and trade-offs involved in the optimization of the network. The use of mathematical programming models on an enterprise-wide scale to address strategic decisions considering various process integration alternatives yielded substantial benefits. These benefits not only materialize in terms of economic considerations, but also in terms of process flexibility and improvements in the understanding of the process interactions and systems limitations. Although the methodology was applied on a network of refineries, it can be readily extended to cover any network of continuous chemical processes. [Copyright &y& Elsevier]

Abstract: This paper describes a distributed continuous simulation of an industrial scale case study using DCOM components. The proposed approach considers each part of the simulation as a DCOM component. These components contain simulations performed with an efficient modelling language, EcosimPro, while other software elements communicate and synchronize the execution of the components. The case study considers an industrial scale simulation corresponding to a beet sugar factory. Several aspects, like the effect of increasing the distribution degree or how it is restricted by the computational requirements of the simulations and the performance of the machines have been studied. The paper describes also an operators training simulator for a sugar factory, based on the distributed simulation. [Copyright &y& Elsevier]

This paper presents a review on the design of batch plants where both the grassroots and the retrofit design problems are analysed. Two types of plants are considered, the multi-product and the multipurpose plants. The characterisation of the design problem is made and the key decisions and elements involved are identified. Limitations and gaps in the existing published approaches are discussed. Although, a considerable number of studies have appeared in the last 15 years there is still a need for further development on the models generalisation and applications. Further attention should be given to the multi-objective design where different objectives are present at the design level. Great investment is also required on the solution methods so as real case problems can be solved in reasonable computational time. The paper concludes with the identification of future research challenges in the design and retrofit of batch plants. [Copyright &y& Elsevier]

Abstract: Informatics infrastructure plays a crucial role in supporting different decision making activities related to pharmaceutical product development, pilot plant and commercial scale manufacturing by streamlining information gathering, data integration, model development and managing all these for easy and timely access and reuse. The foundation of such an infrastructure is the explicitly and formally modeled information. This foundation enables knowledge in different forms, and best manufacturing practices, to be modeled and captured into tools to support the product lifecycle management. This paper discusses the development of ontologies, Semantic Web infrastructure and Web related technologies that make such an infrastructure development possible. While many of the issues addressed in this paper are applicable to a wide spectrum of molecular-based products, we focus our work on the development of pharmaceutical informatics to support Active Pharmaceutical Ingredient (API) as well as drug product development as case studies to illustrate the various aspects of this infrastructure. [Copyright &y& Elsevier]

Abstract: This paper analyses multivariate statistical techniques for identifying and isolating abnormal process behaviour. These techniques include contribution charts and variable reconstructions that relate to the application of principal component analysis (PCA). The analysis reveals firstly that contribution charts produce variable contributions which are linearly dependent and may lead to an incorrect diagnosis, if the number of principal components retained is close to the number of recorded process variables. The analysis secondly yields that variable reconstruction affects the geometry of the PCA decomposition. The paper further introduces an improved variable reconstruction method for identifying multiple sensor and process faults and for isolating their influence upon the recorded process variables. It is shown that this can accommodate the effect of reconstruction, i.e. changes in the covariance matrix of the sensor readings and correctly re-defining the PCA-based monitoring statistics and their confidence limits. [Copyright &y& Elsevier]

Abstract: The microlithography process is the most critical step in the fabrication of nanostructures for integrated circuit manufacturing. The most important variable in the microlithography process is the linewidth or critical dimension (CD), which perhaps is the single variable with the most direct impact on the device speed and performance. In this paper we control one of the key parameters that impact the final CD: the photoresist film thickness. Due to thin film interference, CD varies with photoresist thickness. In the past, fixed controllers were used. However, as different recipes, including different photoresists, may be used for different batches, a fixed controller has to be retuned each time. We proposed the use of self-tuning control – iterative feedback tuning – in this paper and have obtained improvement in both thickness uniformity and convergence time. [Copyright &y& Elsevier]

Abstract: This paper addresses the integrated planning/scheduling of chemical supply chains (SC) with multi-product, multi-echelon distribution networks taking into account financial management issues and suggests a novel approach for enterprise wide management. In order to tackle this problem, it is derived a mathematical formulation combining a scheduling/planning model with a cash flow and budgeting formulation. To motivate the use of such integrated model, a sequential scheme representing traditional enterprise practices is firstly applied. Within this strategy, scheduling and planning decisions are taken firstly, and finances are fitted afterwards considering the cash flows associated to the scheduling/planning decisions previously computed as input parameters. The comparison between the results of the sequential approach and those of the integrated model highlight the advantages of the latter option, in which scheduling/planning and cash management decisions are optimized in unison with a common objective of maximizing the change in equity achieved by the company. The modeling approach developed in this paper and the obtained results suggest that a new conceptual strategy in enterprise management systems consisting of the integration of the financial models of the enterprise with those dealing with the operative area is a must to improve the firm’s performance and its overall earnings and ensuring also healthy cash flow management. [Copyright &y& Elsevier]

Abstract: A method for environmental performance assessment is presented in this paper. The proposed method considers the procedure for environmental performance comparison of design alternatives as a multi-criteria decision-making (MCDM) problem. Integrated assessment model based on analytic hierarchy process (AHP) is presented for solving MCDM problem. An integrated environmental index (IEI) for chemical processes is proposed to combine resource conservation, energy consumption and potential environmental impacts associated with releases. This paper presents IEI as an environmental performance comparison index. As a case study, two alternatives of ethanol production are assessed by means of the proposed method. [Copyright &y& Elsevier]

The kernel of industrial advanced process control (APC) lies in the formulation and solution of model predictive control (MPC) problems, which specify the controller moves according to the solution of an optimal control problem at each sampling time. A significant challenge is the online computation for large-scale industrial systems. As the state-of-the-art APC technology, the Shell–Yokogawa Platform for Advanced Control and Estimation (PACE) has adopted a systematic framework of handling dynamic optimization of large-scale systems, where an automatic decomposition procedure generates subsystems for distributed MPC. The decomposition is implemented on network representations of the MPC models that capture interactions among process variables, with community detection used to maximize the statistical significance of the subnetworks with preferred internal interconnections. This paper introduces the fundamentals of such a decomposition approach and this functionality in PACE, followed by a case study on a crude distillation process to showcase its industrial application. • An industrial implementation of auto-decomposition for distributed MPC is reported. • The platform for advanced control of Shell–Yokogawa, PACE, is introduced. • A directed network captures structural dynamic interactions among process variables. • Community detection algorithm is practically refined to yield desirable subsystems. • Industrial-scale plantwide applications showcase the computational acceleration. [ABSTRACT FROM AUTHOR]

Fault diagnosis (FD) has received attention because of its importance in maintaining safe operations of industrial processes. Recently, modern data-driven FD approaches such as deep learning have shown encouraging performance. Particularly, convolutional neural networks (CNNs) offer an alluring capacity to deal with multivariate time-series data converted into images. Nonetheless, existing CNN techniques focus on capturing local correlations. However, global spatiotemporal correlations often prevail in multivariate time-series data from industrial processes. Hence, extracting global correlations using CNNs from such data requires deep architectures that incur many trainable parameters. This paper proposes a novel local–global scale CNN (LGS-CNN) that directly accounts for local and global correlations. Specifically, the proposed network incorporates local correlations through traditional square kernels and global correlations are collected utilizing spatially separated one-dimensional kernels in a unique arrangement. FD performance on the benchmark Tennessee Eastman process dataset validates the proposed LGS-CNN against CNNs, and other state-of-the-art data-driven FD approaches. • A novel convolutional neural network is proposed for complex fault diagnosis tasks. • The proposed model can handle both local and global spatiotemporal correlations. • The ability to capture global spatiotemporal dependencies is mathematically analyzed. • A closer look at the local receptive fields of the proposed model is provided. • An evaluation against a group of data-driven fault diagnosis techniques is included. [ABSTRACT FROM AUTHOR]

Highlights • Simulation study to economics optimizing control and robust multi-stage optimizing control. • Discussion of numerical aspects of optimizing control as well as the software realization using high-fidelity high-dimensional process models of complex chemical processes. • Control problem formulation for the dynamic product changeover in a multi-product process. • Robustification of economics optimizing control by multi-stage nonlinear model predictive control in case of the presence of model uncertainty. Abstract By the use of online optimization, the profitability of chemical processes can be enhanced while meeting process-related, environmental and ecological constraints. Two well-known techniques to achieve an economically optimal operation of chemical processes are repetitive stationary optimization (RTO) combined with advanced control and direct economics optimizing control on a receding horizon (also called one-layer approach or dynamic RTO). Direct economics optimizing control can react faster to disturbances and is better suited for the optimization of transients between different products. On the other hand it is computationally demanding and requires sufficiently accurate dynamic models. In this paper, we discuss economics optimizing control of a very complex process, a multi-product transesterification reaction that is realized in a reactive distillation process. The process model, the specification of the economics optimizing controller, the implementation of the dynamic optimization, and the robustification of the controller by a multi-stage approach are discussed using simulation studies. [ABSTRACT FROM AUTHOR]

Highlights • A novel piecewise state feedback controller is introduced. • The chemical process optimal control problem is formulated as an optimal control problem of switched dynamical systems. • A gradient-based parallel optimization algorithm is developed for solving the optimal control. • Convergence results indicate that the gradient-based parallel optimization algorithm is global convergent. • Three chemical process optimal control problems are given to illustrate the effectiveness of the proposed algorithm. Abstract This paper considers an optimal control problem of chemical processes with a novel piecewise state-feedback controller. Firstly, the chemical process optimal control problem is formulated as a switched dynamical system optimal control problem, which can be transformed into a parameter optimization problem. Next, to achieve rapid convergence from remote starting points, we propose a novel gradient-based optimization algorithm, which is suitable to a parallel implementation because the step is improved by updating its direction as well as its length simultaneously before moving to the next iteration, and the step computation involves only the inner products of vectors. Then, the convergent properties of the parameter optimization problem to the original optimal control problem are discussed. Finally, the numerical simulation results show that the gradient-based parallel optimization algorithm is an effective alternative method for solving the chemical process optimal control problems. [ABSTRACT FROM AUTHOR]

Value chain integration is an ongoing challenge: while computing power has improved, there is little modeling consistency across the system. This paper bridges this gap by proposing a novel formulation for train scheduling, a linking element of value chains, using the Unit-Operation-Port-State Superstructure (UOPSS). Train scheduling is a challenging problem: rail lines can be hundreds of kilometers long with train crossing strategies that are based on a train station level, while also requiring results with a minute-time scale resolution. In mixed-use rail systems with limited passing loop infrastructure, trains have different passing priorities and lengths, thus differing in their ability to use passing loops. The proposed model is the basis of the Hatch Rail Optimizer (HRO) software. In addition to small case studies, the power of HRO is demonstrated through a practical case study involving a 370 km rail corridor with five different train sizes over a week-long scheduling horizon. [ABSTRACT FROM AUTHOR]

This paper presents a fault detection method based on augmented kernel Mahalanobis distance (AKMD) for monitoring nonlinear dynamic processes. In order to reflect the information of dynamic correlations, the measurements are stacked into augmented vectors at adjacent sampling instants. The augmented kernel Mahalanobis distance serves as the detection index, and its control limit is determined by the empirical method with assigning a significance level. Contrary to the mainstream of process monitoring methods based on principal component analysis (PCA), dimensionality reduction is not used here. The disadvantage of dimensionality reduction and space partition is discussed, and the improvement of fault detectability via data augmentation is analyzed. In addition, the computational complexity of the proposed method is acceptable. For training dataset containing m variables and n samples, if n ≫ m , the online computational burden of the proposed method is about O ( n 2 ) . Simulations about a nonlinear dynamic process and the benchmark Tennessee Eastman process (TEP) both illustrate higher detection rates of the proposed method, compared with conventional multivariate statistical process monitoring (MSPM) methods such as PCA and its variants. [ABSTRACT FROM AUTHOR]

• This paper analyzes the interconnection of fault detection and diagnosis, risk assessment and abnormal situation management. • The paper reviews the existing methods and models related to fault detection and diagnosis, abnormal situation management, and risk. • The paper also present authors' thoughts on the research direction in the area of process safety in industrial 4.0. • This paper aims to serve as a road map for process safety research to enable safer and sustainable development. Industry 4.0 provides substantial opportunities to ensure a safer environment through online monitoring, early detection of faults, and preventing the faults to failures transitions. Decision making is an important step in abnormal situation management. Assigning risk based on the consequences may provide additional information for abnormal situation management decisions to prevent the accident before it occurs. This paper analyzes the interconnections between the three essential aspects of process safety: fault detection and diagnosis (FDD), risk assessment (RA), and abnormal situation management (ASM) in the context of the current and next generation of process systems. The authors present their thoughts on research directions in process safety in Industry 4.0. This article aims to serve as a road map for the next generation of process safety research to enable safer and sustainable process operations and development. [ABSTRACT FROM AUTHOR]

Principal component regression (PCR) has been widely used as a multivariate method for data-based soft sensor design. In order to take advantage of probabilistic features, it has been extended to probabilistic PCR (PPCR). Commonly, industrial processes operate in multiple operating modes. Moreover, in most cases, outputs are measured at a slower rate than inputs, and for each sample of input variable, its corresponding output may not always exist. These two issues have been solved by developing the mixture semi-supervised PPCR (MSPPCR) method. In this paper, we extend this developed model to the case of simultaneous missing data in both input and output. Missing data in multidimensional input space constitutes a significantly more challenging problem. Missing input data occurs frequently in industrial plants because of sensor failure and other problems. We develop and solve the MSPPCR model by using the expectation-maximization (EM) algorithm to deal with missing inputs, in addition to missing outputs and multi-mode conditions. Finally, we present two case studies to demonstrate its performance. [ABSTRACT FROM AUTHOR]

This paper presents a multi-compartment population balance model for wet granulation coupled with DEM (discrete element method) simulations. Methodologies are developed to extract relevant data from the DEM simulations to inform the population balance model. First, compartmental residence times are calculated for the population balance model from DEM. Then, a suitable collision kernel is chosen for the population balance model based on particle–particle collision frequencies extracted from DEM. It is found that the population balance model is able to predict the trends exhibited by the experimental size and porosity distributions by utilising the information provided by the DEM simulations. [ABSTRACT FROM AUTHOR]

• Present a detailed review on manufacturing system maintenance operations in the downstream petroleum industry. • Identify the policies, optimization methodologies, implementation frameworks, and other maintenance efficiency enhancement efforts in the industry to provide general guidance for practitioners and researchers. • Organize maintenance optimization approaches into two broad categories, i.e., based on the criticality of the equipment and based on tools. • Track the recent trends in the maintenance implementation approach, identify gaps, and recommend future research directions to advance the industry's maintenance functions. This paper presents a literature review on maintenance operations in the downstream petroleum industry. This process industry comprises facilities ranging from processing units to distribution networks, which makes the maintenance activities diverse. Maintenance optimization approaches from over 120 articles have been organized into two broad categories. The first category implemented maintenance operations according to the criticality of equipment by applying methods like American Petroleum Institute, Analytical Hierarchy Process, and Failure Modes and Effect Analysis. The second category applied various optimal policies by adopting different tools such as mathematical models (probability, statistics, linear or nonlinear optimization methods, fuzzy logic), heuristic (metaheuristics) algorithms (genetic algorithm, firefly algorithm), data analytics (machine learning), and Internet of Things. The review also included maintenance implementation frameworks, planning & scheduling methods, safety, mechanization, and evaluation procedures. It also tracks the recent trends in the maintenance implementation approach, identifies gaps, and recommends future research directions. [ABSTRACT FROM AUTHOR]

Detrended Fluctuation Analysis (DFA) is a reliable and assumption-free approach for gauging the complexity of a time series. In this paper, an online oscillations detection paradigm is presented, which integrates the potential of DFA in detecting abnormal coherent fluctuations with the Empirical Wavelet Transform (EWT) efficiency in extracting the characteristics of oscillations. However, the standard EWT fails to separate modes oscillating at close frequencies, resulting in an incorrect decomposition. Furthermore, the lack of an appropriate stopping criterion frequently results in the signal being over-decomposed into several inconsequential components. Therefore, owing to the capability of DFA to differentiate between fluctuations stemming from noise and coherent fluctuations arising from genuine oscillations, an Improved EWT (IEWT) is presented to mitigate these issues and accurately extract only compelling oscillating modes. The proposed DFA-based IEWT framework is verified on simulated applications and data from real industrial processes, illustrating its effectiveness. • An integrated monitoring scheme using DFA and IEWT for online oscillations detection. • Oscillations characterisation based on IEWT for industrial control loops. • This method works effectively under different noise density levels. • Incipient oscillations are detected efficiently at their early stage. • The proposed IEWT outperforms the standard EWT in decomposing the signals. [ABSTRACT FROM AUTHOR]

• Surrogate models have been developed for the design of tablet manufacturing. • Random forest regression was chosen for dissolution behavior prediction. • Case studies were performed for both dry and wet granulation methods. • Developed models can reduce the calculation time with high accuracy. • Critical input and intermediate parameters were found by sensitivity analysis. This paper presents surrogate models of dissolution behavior to identify the critical input parameters in tablet manufacturing processes. Dissolution behavior is a critical quality attribute, and is defined as the time profile of fraction of dissolved active pharmaceutical ingredients in a solvent from a tablet. Dissolution behavior was first calculated by mechanistic models using gPROMS® and fitted by the Weibull model. Random forest regression calculated the Weibull model parameters from the input model parameters. The developed models enabled to reduce computational cost without losing accuracy. Case studies were performed for both dry and wet granulation processes using paracetamol. The results identified key input and intermediate parameters to the dissolution behavior, e.g., granule bulk density and porosity. In addition, wet granulation had a tendency to make dissolution faster than dry granulation. The proposed approach can contribute to enhance simulation capabilities in the limited time of pharmaceutical process development. [Display omitted] [ABSTRACT FROM AUTHOR]

The integration of data-driven modeling techniques in machine learning applications, such as multiclass classification, has resulted in robust classifier designs. However, one of the main drawbacks of this approach has been the rising complexity of modeling as the number of classes in the system increases, which may eventually make the overall design of the classifier unfavorable regardless of the expected performance. In this paper, we will discuss the design of a novel logic-based Bayesian-optimized Gaussian process (BOGP) classifier that aims to minimize the number of independent empirical models needed to accurately diagnose multiple distinct fault classes in industrial process. Moreover, the fault classification accuracy of the BOGP classifier is compared to the respective performances of other methods published in literature, and the Tennessee Eastman process is used as a benchmark case study for all methods. • Accurate fault classification is crucial for operational safety and maintenance. • Multi-class systems drive the need for compact and adaptive classifier designs. • Bayesian-optimized Gaussian process classifier proposed to diagnose faults. • Tennessee Eastman industrial process used as benchmark case study. [ABSTRACT FROM AUTHOR]

Optimisation of cell culture processes is an extremely challenging problem for the biomanufacturing industry. Limited data availability, coupled with biological complexity in modelling highly variable living cells, necessitates a decision support methodology that is performant under high levels of uncertainty. Raw materials, experimental & manufacturing facilities, and human expertise are all steeply expensive and availability is tightly constrained — planning their allocation is subject to the core uncertainties underlying the behaviour of living cells. This paper presents a novel decision support methodology for optimisation under high uncertainty. Optimisation techniques require an "objective function" that maps decision variables to the quantity being optimised, so that the decision space can be explored to find an optimum. By learning multiple types of objective function candidates with different levels of fidelity to real-world processes, our method mitigates the risk of picking a poor approximation of the objective function due to sampling effects and algorithmic randomness. Wet lab experimentation on a biomanufacturing feed optimisation problem verified that inferred candidates can successfully support domain experts in designing a new optimised feed strategy with significantly higher product yield than the current industrial control strategy. Our results indicate potential for extending our methodology to the optimisation of other complex industrial processes. • Populations of Objective Functions to mitigate uncertainty. • Machine Learning to Learn objective function. • Optimising Product Yield for manufacturing bioprocesses. • Real world validation. [ABSTRACT FROM AUTHOR]