A critical review of statistical calibration/prediction models handling data inconsistency and model inadequacy

Inference of physical parameters from reference data is a well‐studied problem with many intricacies (inconsistent sets of data due to experimental systematic errors; approximate physical models…). The complexity is further increased when the inferred parameters are used to make predictions—virtual measurements—because parameter uncertainty has to be estimated in addition to parameters best value. The literature is rich in statistical models for the calibration/prediction problem, each having benefits and limitations. We review and evaluate standard and state‐of‐the‐art statistical models in a common Bayesian framework, and test them on synthetic and real datasets of temperature‐dependent viscosity for the calibration of the Lennard‐Jones parameters of a Chapman‐Enskog model. © 2017 American Institute of Chemical Engineers AIChE J, 63: 4642–4665, 2017     

Dynamic modeling and validation of a biomass hydrothermal pretreatment process—a demonstration scale study

Hydrothermal pretreatment of lignocellulosic biomass is a cost effective technology for second generation biorefineries. The process occurs in large horizontal and pressurized thermal reactors where the biomatrix is opened under the action of steam pressure and temperature to expose cellulose for the enzymatic hydrolysis process. Several by‐products are also formed, which disturb and act as inhibitors downstream. The objective of this study is to formulate and validate a large scale hydrothermal pretreatment dynamic model based on mass and energy balances, together with a complex conversion mechanism and kinetics. The study includes a comprehensive sensitivity and uncertainty analysis, with parameter estimation from real‐data in the 178–185°C range. To highlight the application utility of the model, a state estimator for biomass composition is developed. The predictions capture well the dynamic trends of the process, outlining the value of the model for simulation, control design, and optimization for full‐scale applications. © 2015 American Institute of Chemical Engineers AIChE J, 61: 4235–4250, 2015     

Optimization problem with normally distributed uncertain parameters

The issue of chemical process optimization when at the operation stage the design specification should be met with some probability and the control variables can be changed has been considered. A common approach for solving the broad class of optimization problems with normally distributed uncertain parameters were developed. This class includes the one‐stage and two‐stage optimization problems with chance constraints. This approach is based on approximate transformation of chance constraints into deterministic ones. © 2013 American Institute of Chemical Engineers AIChE J, 59: 2471–2484, 2013     

Multistream heat exchangers: Equation‐oriented modeling and flowsheet optimization

Multistream heat exchangers (MHEXs), typically of the plate‐fin or spiral‐wound type, are a key enabler of heat integration in cryogenic processes. Equation‐oriented modeling of MHEXs for flowsheet optimization purposes is challenging, especially when streams undergo phase transformations. Boolean variables are typically used to capture the effect of phase changes, adding considerable difficulty to solving the flowsheet optimization problem. A novel optimization‐oriented MHEX modeling approach that uses a pseudo‐transient approach to rapidly compute stream temperatures without requiring Boolean variables is presented. The model also computes an approximate required heat exchange area to determine the optimal tradeoff between operating and capital expenses. Subsequently, this model is seamlessly integrated in a previously‐introduced pseudo‐transient process modeling and flowsheet optimization framework. Our developments are illustrated with two optimal design case studies, an MHEX representative of air separation operation and a natural gas liquefaction process. © 2015 American Institute of Chemical Engineers AIChE J, 61: 1856–1866, 2015     

Moisture content distribution in semibatch drying processes. I. Constant particle drying rate

Semibatch drying processes include a period of feeding wet particles during drying. This results in different residence times of particles in the dryer and thus generation of a moisture content distribution. Describing these processes in a model that takes this moisture content distribution into account can be complicated. Knowledge of minimum or maximum values of the moisture content distribution is often desired for subsequent process steps or end product quality. This article presents a model that not only describes the overall average moisture content in time during semibatch drying put also gives the evolution of the moisture content distribution in time. A novel solution strategy based on the “method of moments” and “method of characteristics” is presented that solves the resulting ordinary differential equations in a piecewise manner. Simulations for a semibatch fluidized bed drying process give results that are feasible and realistic. © 2012 American Institute of Chemical Engineers AIChE J, 2012     

Analysis of activated sludge process using multivariate statistical tools—a PCA approach

Twelve original physical variables of an activated sludge wastewater treatment system are considered. These cross-correlated variables are transformed in new ones that are not correlated by the use of PCA (principal component analysis), a powerful tool for analysis, monitoring and diagnostics of wastewater treatment processes. Just three principal components explain most of the system total variability (78% of total variance). Thus, the ability to describe the overall characteristics of the process using only three principal components will make the analysis, monitoring and diagnostic of the system easier.

Three groups of variables characterizing the system are detected. The first group identifies variables that represent micro-organisms and inert particulate matter arising from cellular decay, while the second group refers to substrates and flow rate. The third group is related to the pH. Based on these results, the present paper shows how to enlarge the ways of interpreting the characteristics of activated sludge wastewater treatment system.     

Meso‐scale statistical properties of gas–solid flow—a direct numerical simulation (DNS) study

Statistical properties of particles in heterogeneous gas–solid flow were numerically investigated based on the results of a three‐dimensional large‐scale direct numerical simulation (DNS). Strong scale‐dependence and local non‐equilibrium of these properties, especially the particle fluctuating velocity (PFV) or granular temperature, were observed to be related to the effect of meso‐scale structures formed by the compromise in competition between fluid and particle dominated mechanisms. To quantify such effects, the heterogeneous structures were partitioned into a gas‐rich dilute phase and a solid‐rich dense phase according to the particle‐scale voidage defined through the Voronoi tessellation. Non‐equilibrium features, such as the deviation of PFV from Gaussian distribution and anisotropy, were found even in phase‐specific properties. A new distribution function for the PFV well characterizing these features was obtained by fitting the DNS results, which takes a typical bi‐disperse mode, with phase‐specific granular temperatures. The implications of these findings to the kinetic theory of granular flow and traditional continuum models of gas–solid flow were also discussed. © 2016 American Institute of Chemical Engineers AIChE J, 63: 3–14, 2017     

A thermal‐flywheel approach to distributed temperature control in microchannel reactors

Microchannel reactors are a promising route for monetizing distributed natural gas resources. However, intensification and miniaturization represent a significant challenge for reactor control. Focusing on autothermal methane‐steam reforming reactors, a novel microchannel reactor temperature control strategy based on confining a layer of phase‐change material (PCM) between the reactor plates is introduced. Melting‐solidification cycles, which occur with latent heat exchange at constant temperature, allow the PCM layer to act as an energy storage buffer—a “thermal flywheel”—constituting a distributed controller that mitigates temperature excursions caused by fluctuations in feedstock quality. A novel stochastic optimization algorithm for selecting the PCM layer thickness (i.e., distributed controller “tuning”) is introduced. Furthermore, a hierarchical control structure, whereby the PCM layer is complemented by a supervisory controller that addresses persistent disturbances, is proposed. The proposed concepts are illustrated in a comprehensive case study using a detailed two‐dimensional reactor model. © 2013 American Institute of Chemical Engineers AIChE J, 59: 2051–2061, 2013     

A globally convergent method for finding all steady‐state solutions of distillation columns

A globally convergent method is proposed that either returns all solutions to steady‐state models of distillation columns or proves their infeasibility. Initial estimates are not required. The method requires a specific but fairly general block‐sparsity pattern; in return, the computational efforts grow linearly with the number of stages in the column. The well‐known stage‐by‐stage (and the sequential modular) approach also reduces the task of solving high‐dimensional steady‐state models to that of solving a sequence of low‐dimensional ones. Unfortunately, these low‐dimensional systems are extremely sensitive to the initial estimates, so that solving them can be notoriously difficult or even impossible. The proposed algorithm overcomes these numerical difficulties by a new reparameterization technique. The successful solution of a numerically challenging reactive distillation column with seven steady‐states shows the robustness of the method. No published software known to the authors could compute all solutions to this difficult model without expert tuning. © 2013 American Institute of Chemical Engineers AIChE J 60: 410–414, 2014     

Adaptive POD–DEIM basis construction and its application to a nonlinear population balance system

We propose an adaptive algorithm for constructing reduced‐order models (ROMs) of nonlinear systems based on proper orthogonal decomposition (POD) combined with the discrete empirical interpolation method (DEIM). Using an efficient output error estimation, the reduced basis and the DEIM interpolation basis are adaptively adjusted to derive a small, yet accurate ROM. The adaptive algorithm is further explored for a population balance system of a crystallization process. Simulation results show that much smaller and reliable ROMs can be adaptively obtained using the algorithm with ignorable extra computational load as compared with the standard POD–DEIM method. © 2017 American Institute of Chemical Engineers AIChE J, 63: 3832–3844, 2017     

PermSMBR—A new hybrid technology: Application on green solvent and biofuel production

A new technology, the simulated moving bed membrane reactor (PermSMBR), was presented and applied for the production of the green solvent ethyl lactate and of the biofuel 1,1‐diethoxyethane. Its conception was a result of process reintensification for oxygenates production, by integrating the simulated moving bed reactor with hydrophilic membranes to enhance the water removal, leading to high process performance. For ethyl lactate synthesis, the PermSMBR technology proved to have better performance than the reactive distillation (RD) and the simulated moving bed reactor (SMBR) processes; the RD and SMBR processes require more 152 and 165% of ethanol consumption than the new technology, respectively. For the 1,1‐diethoxyethane production, the PermSMBR also leads to a decrease in ethanol consumption of 69% and a productivity enhancement of 53%, when comparing with the SMBR. © 2010 American Institute of Chemical Engineers AIChE J, 57: 000–000, 2011     

Numerical simulation of particle breakage in dry impact pulverizer

A novel method to simultaneously simulate particle motion and its breakage in a dry impact pulverizer was developed. The motion of particles in the pulverizer was calculated using a discrete phase model (DPM)‐computational fluid dynamics (CFD) coupling model. When the particle impacts against a vessel wall, impact stress acting on the particle is calculated from Hertz's theory as a function of the impact velocity. At the same time, the particle strength as a function of the particle size is calculated from Griffith's theory. If the impact stress is larger than the particle strength, the particle is broken and replaced with smaller fragments. The size distribution of the fragments is obtained from a breakage function proposed. The motion of the fragments is calculated again by using the DPM‐CFD coupling model. By repeating the above calculations over the whole particles, the grinding phenomenon can be simulated. The calculated results showed good agreement with the experimental one, and validity of the proposed method was confirmed. © 2013 American Institute of Chemical Engineers AIChE J, 59: 3601–3611, 2013     

Simulation,model‐reduction,and state estimation of a two‐component coagulation process

The issue of state estimation of an aggregation process through (1) using model reduction to obtain a tractable approximation of the governing dynamics and (2) designing a fast moving‐horizon estimator for the reduced‐order model is addressed. The method of moments is first used to reduce the governing integro‐differential equation down to a nonlinear ordinary differential equation. This reduced‐order model is then simulated for both batch and continuous processes and the results are shown to agree with constant Number Monte Carlo simulation results of the original model. Next, the states of the reduced order model are estimated in a moving horizon estimation approach. For this purpose, Carleman linearization is first employed and the nonlinear system is represented in a bilinear form. This representation lessens the computation burden of the estimation problem by allowing for analytical solution of the state variables as well as sensitivities with respect to decision variables. © 2016 American Institute of Chemical Engineers AIChE J, 62: 1557–1567, 2016     

Dual mechanism model for fluid particle breakup in the entire turbulent spectrum

This work provides an in-depth understanding of different breakup mechanisms for fluid particles in turbulent flows. All the disruptive and cohesive stresses are considered for the entire turbulent energy spectrum and their contributions to the breakup are evaluated. A new modeling framework is presented that bridges across turbulent subranges. The model entails different mechanisms for breakup by abandoning the classical limitation of inertial models. The predictions are validated with experiments encompassing both breakup regimes for droplets stabilized by internal viscosity and interfacial tension down to the micrometer length scale, which covers both the inertial and dissipation subranges. The model performance ensures the reliability of the framework, which involves different mechanisms. It retains the breakup rate for inertial models, improves the predictions for the transition region from inertia to dissipation, and bridges seamlessly to Kolmogorov-sized droplets.     

Large‐scale optimization strategies for pressure swing adsorption cycle synthesis

Pressure swing adsorption (PSA) is an efficient method for gas separation and is a potential candidate for carbon dioxide (CO₂) capture from power plants. However, few PSA cycles have been designed for this purpose; the optimal design and operation of PSA cycles for CO₂ capture, as well as other systems, remains a very challenging task. In this study, we present a systematic optimization‐based formulation for the synthesis and design of novel PSA cycles for CO₂ capture in IGCC power plants, which can simultaneously produce hydrogen (H₂) and CO₂ at high purity and high recovery. Here, we apply a superstructure‐based approach to simultaneously determine optimal cycle configurations and design parameters for PSA units. This approach combines automatic differentiation, efficient ODE solvers for the state and sensitivity equations of the PSA model, and state of the art nonlinear programming solvers. Three optimization models are proposed, and two PSA case studies are considered. The first case study considers a binary separation of H₂ and CO₂ at high purity, where specific energy is minimized, whereas the second case study considers a larger five component separation. © 2012 American Institute of Chemical Engineers AIChE J, 58: 3777–3791, 2012     

Simple model to predict gel formation in olefin‐diene copolymerizations catalyzed by constrained‐geometry complexes

We have developed an analytical model to predict the onset of gel formation in ethylene/1‐octene/1,9‐decadiene terpolymerizations using constrained‐geometry catalysts. The model relies on three kinetic parameters to characterize the catalyst response. Polymer resins have been synthesized in a continuous stirred‐tank reactor to determine the model parameters, and to validate the model predictions for polymer properties and for the onset of gel formation and reactor fouling. The experimental results indicate that the free double bonds in 1,9‐decadiene are as reactive as those found in 1‐octene, and that the reactivity of 1,9‐decadiene double bonds decreases after the 1,9‐decadiene molecules become part of a polymer chain. The model predictions of polymer properties agree well with chromatographic, density, and mass‐balance data. Moreover, the model was successful in preventing unintended reactor fouling during the duration of the experimental campaign. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

Concentration polarization in a narrow reverse osmosis membrane channel

Concentration polarization in a narrow reverse osmosis channel is bounded by the channel height and under the influence of variable transverse velocity. An attempt was made in this article to quantify concentration polarization in such a narrow membrane channel. The transverse velocity in the membrane channel was first determined and its impact on concentration polarization was investigated. Based on the concept of retained salt, analytical equations were developed for the wall salt concentration at an arbitrary point in the narrow membrane channel. Finally, development of concentration polarization in typical reverse osmosis channels under various conditions was numerically simulated and discussed. Interesting results on the details of concentration polarization in the narrow reverse osmosis channel that had never been reported before were revealed with this mechanistic model. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

基于加权统计特征KICA的故障检测与诊断方法

针对核独立元分析（kernel independent component analysis, KICA）在非线性动态过程中对微小故障检测率低的问题,提出一种基于加权统计特征KICA（weighted statistical feature KICA, WSFKICA）的故障检测与诊断方法。首先,利用KICA从原始数据中捕获独立元数据和残差数据;然后,通过加权统计特征和滑动窗口获取改进统计特征数据集,并由此数据集构建统计量进行故障检测;最后,利用基于变量贡献图的方法进行过程故障诊断。与传统KICA统计量相比,所提方法的统计量对非线性动态过程中的微小故障具有更高的故障检测性能。应用该方法对一个数值例子和田纳西-伊斯曼（Tennessee-Eastman, TE）过程进行仿真测试,仿真结果显示出所提方法相对于独立元分析（ICA）、KICA、核主成分分析（kernel principal component analysis, KPCA）和统计局部核主成分分析（statistical local kernel principal component analysis, SLKPCA）检测的优势。     

Experimental methods in chemical engineering: Artificial neural networks–ANNs

Artificial neural networks (ANNs) are one of the most powerful and versatile tools provided by artificial intelligence and they have now been exploited by chemical engineers for several decades in countless applications. ANNs are computational tools providing a minimalistic mathematical model of neural functions. Coupled with raw data and a learning algorithm, they can be applied to tasks such as modelling, classification, and prediction. Recently, their popularity has grown remarkably and they now constitute one of the most relevant research areas within the fields of artificial intelligence and machine learning. ANNs are large collections of simple classifiers called neurons. Chemical engineers apply them to model complex relationships, predict reactor performance, and to automate process controllers. ANNs can leverage their ability to learn and exploit large data sets, but they can also get stuck in local minima or overfit and are difficult to reverse engineer. In 2016 and 2017, ANNs were cited in 13 245 Web of Science (WoS) articles, 538 of which were in chemical engineering; the top WoS categories were electrical & electronic engineering (1615 occurrences) artificial intelligence (1253), and energy & fuels (980). The top 4 journals mentioning ANNs were Neural Computing & Applications (117), Neurocomputing (84), Energies (76), and Renewable & Sustainable Energy Reviews (76). In the near future, as larger data sets become available (and arduous to analyze), chemical engineers will be able to apply and leverage more sophisticated ANN architectures.     

On reducing false alarms in multivariate statistical process control

The primary objective of this note is to reduce the false alarms in multivariate statistical process control (MSPC). The issue of false alarms is inherent within MSPC as a result of the definition of control limits. It has been observed that under normal operating conditions, the occurrence of “out-of-control” data, i.e. false alarms, conforms to a Bernoulli distribution. Therefore, this issue can be formally addressed by developing a Binomial distribution for the number of “out-of-control” data points within a given time window, and a second-level control limit can be established to reduce the false alarms. This statistical approach is further extended to consider the combination of multiple control charts. The proposed methodology is demonstrated through its application to the monitoring of a benchmark simulated chemical process, and it is observed to effectively reduce the false alarms whilst retaining the capability of detecting process faults.