Computers & Chemical Engineering最新文献

Climate change, pandemics, and economic crises have created complex challenges for supply chains. Managing such situations requires the development of reliable decision-making frameworks. In this paper, a multi-level, multi-product, and multi-period closed-loop supply chain is studied with environmental considerations. A bi-objective mixed-integer linear programming model is presented for facility location, flow allocation, and transportation mode determination. The objectives of the model are to minimize the total cost and maximize the reliability of suppliers to meet the needs of factories. In the area of reliability engineering, a new approach is defined for modeling the probability of supplier availability considering catastrophic failures caused by pandemics, economic sanctions, and other failure modes. Furthermore, the decision-maker can handle the emission of greenhouse gases by an upper-bound constraint. In order to face the simultaneous uncertainty of demand and the maximum CO₂ emission allowed, a scenario-based two-stage stochastic programming approach is proposed. The improved version of the augmented ε-constraint method, known as AUGMECON2, is used to solve the proposed model. The efficiency of the model and the proposed solution approach are investigated through a real-world case study of a battery manufacturing company in Iran.

Chemical compound classification, toxicity prediction, and environmental risk assessments are critically important in various applications within the field of chemistry. Deep learning models provide highly effective tools for extracting features from complex large datasets and performing classification tasks. Four different deep learning models, namely ResNet50V2, VGG19, InceptionV3, and MobileNetV2, have been compared with the random forest (RF) and k-nearest neighbors (KNN) algorithms. The results obtained from experiments conducted using QRCODE images of the Tox21SMILES dataset demonstrate the effectiveness of deep learning models for classifying chemical compounds and showcase the performance of different classification algorithms. The findings of the study thoroughly evaluate the performance of deep learning models and classification algorithms in the task of chemical classification. While ResNet50V2 and VGG19 models achieve high accuracy and precision, InceptionV3 and MobileNetV2 models provide more balanced results. Additionally, in terms of classification algorithms, the k-nearest neighbors (KNN) algorithm generally outperforms the Random Forest (RF) algorithm. Although the RF algorithm achieves good accuracy, the KNN algorithm proves to be more effective in terms of sensitivity and F1 score. These results emphasize the factors to consider when choosing which deep learning model or classification algorithm to use in chemical classification tasks. In conclusion, this study presents a comprehensive analysis comparing the performance of deep learning models and classification algorithms in chemical classification tasks. The selection of the most suitable model and algorithm for a specific task supports achieving better results in the classification of chemical compounds and related applications.

Emerging chemical technologies can upcycle plastic waste by producing high-value polymers and other products. In this work, we study the economic and environmental benefits of deploying an upcycling infrastructure in the continental United States for producing low-density polyethylene (LDPE) and polypropylene (PP) from post-consumer plastic waste. Our analysis is based on a computational framework that integrates techno-economic analysis, life-cycle assessment, and value chain optimization. Our results demonstrate that the infrastructure could generate a market of nearly 20 billion USD per year and that this market is robust to various externalities. Our analysis also indicates that the infrastructure can achieve a plastic-to-plastic degree of circularity of 34% relative to residential plastic waste production, and leads to significant environmental benefits over alternative waste disposal methods, including 69%–75% lower greenhouse gas emissions than waste-to-energy systems and 38 million tonnes of avoided landfill waste per year.

Data-driven modeling is essential in chemical engineering, especially in complex systems like wastewater treatment plants. Recurrent neural networks are effective for modeling parameters in wastewater treatment process such as dissolved oxygen concentration and chemical oxygen demand due to their nonlinear adaptability. However, traditional models face challenges such as the requirement for larger datasets and more frequent sampling, noisy measurements, and overfitting. To address this, physics-informed neural networks integrate physical knowledge for improved performance. In our study, we apply both approaches to a wastewater treatment plant, enhancing prediction performance. Our results demonstrate that physics-informed models perform successfully in offline and online validation, especially when standard methods fail. They maintain effectiveness without frequent updates. Yet, integrating physics-informed knowledge can introduce noise when standard methods suffice. This result points out the need for careful consideration of model choice in different scenarios.

Surrogate modeling is used to replace computationally expensive simulations. Neural networks have been widely applied as surrogate models that enable efficient evaluations over complex physical systems. Despite this, neural networks are data-driven models and devoid of any physics. The incorporation of physics into neural networks can improve generalization and data efficiency. The physics-informed neural network (PINN) is an approach to leverage known physical constraints present in the data, but it cannot strictly satisfy them in the predictions. This work proposes a novel physics-informed neural network, KKT-hPINN, which rigorously guarantees hard linear equality constraints through projection layers derived from KKT conditions. Numerical experiments on Aspen models of a continuous stirred-tank reactor (CSTR) unit, an extractive distillation subsystem, and a chemical plant demonstrate that this model can further enhance the prediction accuracy.

Technologies based on Artificial Intelligence (AI) and Machine Learning (ML) concepts are advancing at a rapid pace. The new paradigms are challenging the status-quo of mature automation and control technologies in industry. Autonomous operation is a frequently stated goal of AI evangelists and technology providers. This white paper gives an overview of the current state of art of advanced process control and optimization technologies. It also provides a brief summary of the AI and ML-based approaches that address the closed-loop control and optimization space. Some results from industrial implementations are shared for both conventional and AI/ML-based approaches. Experience from four industrial applications is shared, covering rigorous model-based, machine learning and hybrid approaches to real-time optimization and control problems. The applications range from unit-based control & optimization to refinery & network wide optimization. A set of high-level requirements that need to be satisfied regardless of the underlying technology for closed-loop autonomous operations is reviewed. The article concludes with some future directions and perspectives highlighting areas where the emerging technologies may have significant impact in industry.

Chemometric models for multicomponent mixture analysis are reliant upon representative and accurate training data. However, the required amount of training data can increase exponentially with the number of constituents. Shifting spectral baselines and changing process objectives may necessitate recurring calibration data collection. Furthermore, all possible constituents may not be known prior to analysis. Here, we introduce a preprocessing procedure that reduces the burden of collecting extensive calibration datasets by filtering out non-target species (species that are not part of the calibration data) in real-time. The method, nonnegatively constrained classical least squares (NCCLS), utilizes a spectral nonnegativity constraint on non-target species to remove them from mixture spectra. The preprocessing method is physically motivated, does not rely on time-series data, and can operate in real-time. NCCLS is compared to established methods using in silico data and an experimental dataset comprised of Raman spectra and attenuated total reflectance - Fourier transform infrared spectra.

This study proposes a novel fault detection and identification method using Continuous Wavelet Transform (CWT) and a three-dimensional Convolutional Neural Network (3D-CNN). In particular, multivariate time series data from chemical plants were divided by a time-shifting window and transformed into scalograms using CWT. These scalograms were fed to 3D-CNN to generate outputs indicating the faults that occurred in the process. We applied the proposed method to a Tennessee Eastman process dataset. The proposed method adequately captures the characteristics in the time–frequency domain and exhibited good fault detection and identification performances on the dataset.

Model-based design of experiments (MBDoE) is a powerful framework for selecting and calibrating science-based mathematical models from data. This work extends popular MBDoE workflows by proposing a convex mixed integer (non)linear programming (MINLP) problem to optimize the selection of measurements. The solver MindtPy is modified to support calculating the D-optimality objective and its gradient via an external package, SciPy, using the grey-box module in Pyomo. The new approach is demonstrated in two case studies: estimating highly correlated kinetics from a batch reactor and estimating transport parameters in a large-scale rotary packed bed for CO${}_2$ capture. Both case studies show how examining the Pareto-optimal trade-offs between information content measured by A- and D-optimality versus measurement budget offers practical guidance for selecting measurements for scientific experiments.

In chemical process industry, the input-output data of processes display complex nonlinear dynamics and strong influences of unobserved hidden states. Their behavior must be modeled using nonlinear time series with an observer-predictor structure. To address this, a sequence-to-sequence model with a memory layer was proposed for a high-density polyethylene slurry reactor. The memory layer retains the chronological contribution of the observer and predictor and effectively captures the lengthy time response. A physics-guided approach was adopted to ensure the directional consistency between input and output variables in key control loops. In this way, a deep learning model can be obtained with historical data and there is no need for plant tests. The resulting model can be used in a nonlinear model predictive control system that not only quickly navigates different grade transitions but also provides steady-state control even though key internal state variables such as catalyst activity change.