Chemical Engineering Science最新文献

Data-driven soft sensors have become popular tools for estimating critical quality variables in the process industry. However, in practical applications, it is very common that the unlabeled data are abundant but the labeled data are scarce, which poses a great challenge for developing high-performance data-based soft sensors. Thus, a dynamic dimensionality reduction-assisted large-scale pseudo label optimization method (DDR-LSPLO) is proposed for achieving sample expansion. This method repeatedly converts the LSPLO issue into a reduced-dimension pseudo label optimization problem with the low-confidence pseudo labels as new decision variables during the evolutionary optimization process. Meanwhile, to tackle the sample imbalance problem resulting from the inclusion of large-scale pseudo-labeled samples, a sample expansion and weighting-based quality-relevant autoencoder (SEWQAE) is developed for semi-supervised soft sensor modeling. The effectiveness and superiority of the proposed DDR-LSPLO and SEWQAE methods are verified through an industrial chlortetracycline (CTC) fermentation process and a simulated Tennessee Eastman (TE) chemical process.

Layered δ-MnO₂ prepared by a one-step redox method was shown to be deactivated during oxidation of formaldehyde at room temperature. Recovery of the formaldehyde degradation activity was investigated after thermal regeneration at different temperatures. XRD, SEM, TEM, H₂-TPR, XPS, TGA and DRIFTS characterization were used to analyze the physical properties of fresh, deactivated and thermally regenerated catalysts. The results showed that deactivation of catalyst was caused by less oxygen vacancies due to increased Mn⁴⁺ content during formaldehyde degradation and formation of formate blocking active sites. Thermal regeneration helped to decompose formate at the catalyst surface, restoring some of the active sites. Interplanar spacing of MnO₂ became wider and the number of Mn³⁺ on exposed crystal faces increased. More oxygen vacancies were formed. The activity of deactivated catalyst was restored. Formaldehyde degradation rate of catalyst regenerated at 200 °C remained above 80 % after 6 h, demonstrating the possibility of waste layered catalyst recycling.

Many continuously operated pharmaceutical process routes have been presented recently. Most of these cover the synthesis of the active pharmaceutical ingredient (API) or solid dosage processing. However, the API purification is also gaining attraction. One widespread and waste-intensive unit operation for purification is the liquid–liquid-extraction (LLE). In continuous manufacturing active process control is required, especially for fast processes. Control concepts must be able to react on product quality deviations caused by disturbances or inadequate process settings in real-time. In this study different control concepts for LLE were developed for an extraction column and a multistage extraction. A universally applicable process model was derived and parametrized. The impact of several control concepts including different real-time measurements was evaluated in simulation for both LLE process routes. The results show that simulation tools based on proper process models can support the selection of the most efficient process route and of suitable control concepts.

In this study, we have developed a mathematical model for a three-phase separator. The model consists of two sections: the inlet section and the separation section, separated by a perforated calming baffle. In the inlet section, two dispersion layers undergo droplet size evolution due to turbulent breakage and coalescence, described by a spatially homogeneous PBE. In the separation section, the two dispersion layers flow alongside each other and interact at an interface. The volumetric flow and velocity profiles are influenced by interfacial coalescence, with considerations for plug and laminar flow assumptions. The model incorporates droplet gravity-driven transport using the Kumar and Hartland model, binary and interfacial coalescence employing a film drainage model, and an effective diffusion term to account for the formation of the dense packed layer which ensures a physical volume fraction range of 0–1. Steady-state and transient numerical solvers are developed to solve the resulting advection–diffusion equations. Additionally, a series of experiments were conducted using a lab-scale multi-parallel pipes separator to investigate the impact of varying volume fractions and flow rates on the separation efficiency of the equipment. The model results are compared with the experimental data which shows relatively good agreement.

Multiphysics engineering has been a crucial task in a chemical reactor because complicated interactions among fluid mechanics, chemical reactions, and transport phenomena greatly affect the performance of a chemical reactor. Recently, physics-informed neural networks (PINN) have been successfully applied to various engineering problems thanks to their domain generalization ability. Herein, we introduce a novel application of PINN to multiphysics in a chemical reactor. Specifically, we examined the effectiveness of PINN to reconstruct and extrapolate ethylene conversion in a polymerization reactor. We ran CFD for the polymerization reactor to use in the training process; thereafter, we constructed the PINN by combining the loss of conventional neural networks (NN) with the residuals of the continuity, Navier-Stokes, and species transport physics equations. Our results showed that the PINN more accurately predicted the overall ethylene concentration profile, which is the primary result of multiphysics in the reactor; PINN showed 18 % lower mean absolute error (0.1028 mol/L) than NN (0.1267 mol/L). Furthermore, the PINN satisfactorily predicted the conversion concaveness effect, which is a unique multiphysical effect in a radical polymerization reactor, while NN couldn’t. These results highlight that multiphysics in a chemical reactor may be efficiently predicted and even extrapolated by harnessing physics in neural networks.

This study enhanced the adsorptive capacity of polystyrene (PS) by infusing reduced graphene oxide (rGO) nanoparticles obtained from the synthesis of graphene oxide to produce PS-rGO composites via electrospinning method. Physicochemical characterization of as-synthesized rGO and PS-rGO were carried out through scanning electron microscopy, N₂ physisorption among others. Oil sorption performance of synthesized rGO in crude oil, vegetable oil, fresh engine oil and used engine oil are 130.96 g/g, 121.77 g/g, 105.01 g/g and 100.56 g/g. Oil sorption capacities of electrospun pure PS in crude oil, vegetable oil, fresh engine oil and used engine oil were 46.32 g/g, 38.54 g/g, 35.14 g/g and 32.57 g/g and those of PS-rGO infused with 4 wt% of rGO were found to be 105.52 g/g, 98.86 g/g, 86.25 g/g and 83.47 g/g for crude oil, vegetable oil, fresh engine oil and used engine oil samples respectively. Pseudo second order (PSO) kinetic model fits the sorption data of the four oil samples on the four composite sorbents produced. Intra-particle diffusion (IPD) model evidently showed that sorption of the four oil samples on the four composite sorbents, occurred in three (3) phases. Composites demonstrate high oil adsorption capacity, and are reusable upto three sorption–desorption cycles.

Computation of the critical point of complex fluid mixtures is an important part of understanding their thermodynamic phase behaviour. While algorithms for these calculations are well established, they are often slow when the number of constituting components is large. In this work, we propose a new procedure to significantly accelerate critical point calculations by leveraging deep neural network (DNN) models. A DNN model for critical point predictions of a given mixture is first trained based on the critical points of such a mixture with various compositions. The predictions of the DNN model are then used to initialize both of the commonly used algorithms for mixture critical point calculations: root finding and global minimization. We demonstrate that when using the DNN-based predictions to initialize the root-finding-based and optimization-based algorithms, we can achieve 50-90% and 80-90% reductions in the number of required iterations, respectively.

The use of agricultural and forestry waste to produce functional materials is a significant approach to achieving carbon neutrality. Herein, a green and cost-effective pre-oxidation and self-activation approach has been adapted to produce porous carbon from discarded tobacco stems for supercapacitors. The analysis of tobacco stem structure evolution reveals that the pre-oxidation process facilitated the cross-linked structure of the tobacco stem and the formation of KCl crystals, endowing tobacco stem-derived porous carbon with abundant micropores and high oxygen content during self-activation. The impact of pre-oxidation and self-activation temperature on the carbon structural characteristics of tobacco stems is systematically investigated. The optimized porous carbon exhibited a specific capacitance of 320 F/g at 0.5 A/g with good rate capability. Besides, it delivered a high energy density of 10.68 Wh/kg in a symmetrical supercapacitor. This work provides a green route for preparing carbon electrode materials for high-performance supercapacitors using agricultural and forestry wastes.

Electrocatalytic nitrate reduction reaction (NO₃RR) provides an alternative to the conventional Haber-Bosch process for ammonia synthesis and is an effective method for removal of nitrate ions from polluted waters, which is highly significant from both energy and environmental perspectives. However, NO₃RR involves the complex eight-electron process alongside various nitrogen-containing intermediates and is also in competition with hydrogen evolution reaction, thus demanding highly active and selective electrocatalysts. In this work we prepare a Ni-doped Fe₂O₃ electrocatalyst via a solvent-free route. It is found that the addition of Ni induces the crystalline-phase transformation of Fe₂O₃ from γ-Fe₂O₃ to α-Fe₂O₃. The density functional theory (DFT) results reveal that compared to γ-Fe₂O₃, α-Fe₂O₃ gives rise to a lower potential-determining step (PDS) energy barrier, leading to the more thermodynamically favourable reaction. By modulating the crystalline phase, the optimal catalyst achieves high ammonia yield rates of > 5000 μg h⁻¹ cm⁻² and faradaic efficiencies of > 90 %, showcasing its high electrocatalytic activity and selectivity. From this perspective, this paper provides new insights and strategies for the green nitrate-to-ammonia conversion.

Shale gas is a low-carbon, clean, and high-reserve natural gas resource, but the development process requires a large amount of fresh water and chemicals, which can lead to a large amount of As3+ in the shale gas raw water. The removal of As³⁺ from shale gas raw water is necessary because of the serious hazards that As³⁺ can cause once it enters the human body. In this study, a loofah biocarbon material (CBMM) co-modified by Cucurbit[7]uril (CB[7]) and Fe₃O₄ was prepared. The successful synthesis of the materials was verified by various characterization methods. The material possesses excellent magnetic separation properties and can achieve rapid recovery within 50 s. The adsorption process is spontaneous and endothermic, and the experimental data have excellent correlation with pseudo-first-order kinetic (R² > 0.99) and Langmuir model (R² > 0.99). The maximum adsorption capacity of CBMM was 76.43 mg/g at 20 °C. In addition, CBMM still possessed 74.8 % of the initial adsorption capacity after 7 cycles of the experiment. CBMM also had excellent As³⁺ removal efficiency (90.1 %) in the study of actual shale gas raw water. In conclusion, CBMM is a very promising adsorbent for the removal of As³⁺ from shale gas raw water.