Volume 20, Issue 6

In the steel industry, the propensity of byproduct gas flow fluctuation is a critical factor in energy scheduling. The financial success of the steel industry will greatly benefit from an accurate forecast of its future tendencies. Manufacturers began to focus more on the needs of the client than on the product. Energy-intensive businesses like steel manufacturing are also affected by it. It might be difficult for mass customization firms to determine a product’s overall cost with accuracy. Moreover, energy accounts for 20 %–40 % of the expenditures associated with steel goods. Growing the selection of products may inevitably result in a loss of sustainability. This study investigates the prediction of power consumption in the steel industry by employing Linear Discriminant Analysis Classification (LDAC) and Decision Tree Classification (DTC). To improve the predictive accuracy of these foundational models, two advanced optimizers were utilized: The Electrostatic Discharge Algorithm (EDA) and the Political Optimizer (PO). These optimizers were systematically integrated with the base models to develop innovative hybrid models. Specifically, the LDAC model coupled with EDA forms the LDED framework, the LDAC model combined with PO constitutes the LDPO framework, the DTC model integrated with EDA results in the DTED framework, and the DTC model combined with PO creates the DTPO framework. The DTPO model performs exceptionally well in the Accuracy metric value at the Test section (value of 0.931), while the DTC and DTED models perform similarly well (both with a value of 0.911). In the Precision metric value at the Test section (value of 0.932), the DTPO model performs the best, and the LDED model performs the worst (value of 0.857).

Gasification modeling involves developing and evaluating computational and mathematical frameworks to replicate and forecast the behavior of gasification processes. Among these frameworks, the Decision Tree (DT) model is particularly popular for its capacity to handle both categorical and numerical data while maintaining interpretability. This characteristic shed light on the key factors affecting gasification outcomes. This hybrid strategy harnessed the advantages of DTs for predictive modeling, along with enhancement methods to refine the gasification process. According to the performance metrics, the DTAO hybrid model consistently excelled compared to others, recording the highest R 2 values of 0.989 (CO) and 0.985 (CO 2 ), the lowest RMSE values of 0.996 (CO) and 0.797 (CO 2 ), as well as the lowest OBJ values of 0.935 (CO) and 0.793 (CO 2 ). This confirms its exceptional accuracy and reliability in gasification prediction.

In chemical, petrochemical, and biochemical sectors, bubble columns are essential for facilitating critical reactions like methanation, oxidation, and hydrogenation. Gas hold-up is one of the most important operational parameters that has a significant impact on the performance and efficiency of these multiphase reactions. In order to improve the design and operational efficiency of bubble columns, it is imperative that the gas hold-up be precisely estimated. Creating a sophisticated predictive model with machine learning techniques is a promising solution for obtaining accurate and dependable estimations because of the intricate relationships and the unpredictable impact of numerous factors on gas hold-up. Therefore, this study explores the efficiency of four deep learning models, including the Recurrent Neural Network, Convolutional Neural Network, Long Short-Term Memory, and Gated Recurrent Unit in gas hold-up estimation. This study utilizes the Harris Hawk Optimization algorithm for hyperparameter fine-tuning and enhancing the model’s performance. The optimized Recurrent Neural Network model achieves markable results, attaining a test correlation coefficient of 0.995, root mean square error of 0.005 alongside a mean absolute error of 0.0054. Additionally, this study demonstrate that liquid height is the most influential variable in controlling gas hold-up by sensitivity indices approaching 1, which suggests that even slight variations in liquid height can lead to significant changes in gas retention.

This study addresses the urgent need to reduce environmental impacts in the construction industry by improving the prediction of CO 2 emissions and energy consumption during the production of High-Performance Concrete (HPC). To tackle this challenge, we employed three machine learning (ML) regression models: K-Nearest Neighbor Regression (KNNR), Gradient Boosting Regression (GBR), and eXtreme Gradient Boosting Regression (XGBR) to examine the complex patterns influencing sustainability metrics in HPC manufacturing. Each model was further enhanced using the Fitness Dependent Optimizer (FDO) to increase predictive accuracy and decrease error rates. We evaluated model performance through training, validation, and testing phases, focusing on two objectives: CO 2 emissions and energy usage. Among all the approaches tested, the KNNR and its hybrid version (KNFD) achieved the lowest Weighted Absolute Percentage Error (WAPE) of 0.002 for the energy prediction task. The XGBR and XGFD models followed close behind with WAPEs of 0.05 and 0.04, respectively, outperforming the GBR and GBFD models. Feature selection based on the F-statistic identified “cement” as the most crucial variable among the eight input features. These findings demonstrate the effectiveness of FDO-enhanced regression models in delivering accurate, low-error predictions, which are essential for sustainability-oriented initiatives in construction. This research establishes a solid foundation for data-driven strategies aimed at reducing environmental impacts in HPC production.

High water quality (WQ) is of the essence to human health and environmental sustainability. The resultant effects are viewed in industrial and agricultural productivity and the drinking water supply. In light of the recent trend of water-based tourism and recreational activities, it is now more than ever relevant to commit to stringent enforcement of the regulations on WQ. Compliance with these norms will ensure that the security of drinking water, agricultural productivity, and industrial processes is maintained accordingly. It can predict water quality by finding patterns within different water parameters by analyzing past and present data. The given study explains the use of Support Vector Classification (SVC), Gradient Boost Classification (GBC), and Random Forest Classification (RFC) that have been optimized using Fitness Dependent Optimizer Algorithm (FDOA) and Cheetah Optimization Algorithm (COA), respectively, to improve their accuracy in projection. Considering the results of evaluation metrics for models, the GBCO, SVCO, and RFCO models registered 0.992, 0.966, and 0.944 values of Accuracy, while the GBC, SVC, and RFC single models achieved 0.956, 0.933, and 0.904 values. Implementing the optimization algorithms, especially the COA, enhanced the Accuracy of single models as 3.6 %, 3.3 %, and 4 % for GBC, SVC, and RFC.

To address industrial challenges in thermal efficiency equipment, current research emphasizes optimizing system designs and employing efficient heat transfer fluids. In this paper, the turbulent heat transfer performance of nanofluids in a TG-type vortex generator is investigated by numerical simulations using the mixture model at Re  = 2,600–17,000 operating conditions. Gap-Diameter ratio ( G /D) are investigated in order to select a vortex generator structure with the optimal performance (Best performance when G /D = 1.2). Then the impact of nanofluid type, concentration, and CNT diameter on heat transfer was quantified. It was found that the Nusselt number of Vortex generators was by 1.75–2.44 times higher than that of the empty tube. The nanofluids Al 2 O 3 –H 2 O, SiC–H 2 O and CNT–H 2 O achieved Nu avg increases of 1.04–1.17, 1.30–1.46 and 5.36–5.98 times over water, respectively. By increasing the Fc and suppressing the S T , CNT can reduce total irreversible losses while achieving efficient heat transfer. In addition, the comprehensive heat transfer performance of single-walled carbon nanotubes (SWCNT) is better than that of multi-walled carbon nanotubes (MWCNT). MWCNT = 50 nm, MWCNT < 10 nm and SWCNT exhibited PEC enhancements of 4.03–5.07, 4.48–5.70, and 6.92–9.07 times over water, respectively. This study provides valuable insight into the optimization of heat exchangers and other industrial thermal systems.

The presented study introduces an advanced predictive modeling framework designed to predict CH 4 and C 2 H n yields in the gasification process, in which a method is traditionally employed to convert carbonaceous feedstocks into industrially valuable gases. Accurate prediction of gas yields is essential for improving operational efficiency and promoting environmental sustainability. The proposed approach seeks to advance gasification through machine learning models that suggest improved efficiency, cost-effectiveness, and precision. Such models support real-time decision-making and adaptability across diverse operational conditions, while also elevating safety by minimizing dependencies on conventional experimentation and expanding the scope of controllable parameters. The Naïve Bayes algorithm is employed to model complex relationships between influential input variables and gas production outcomes. To further boost predictive accuracy, the NB model is integrated with two metaheuristic optimizers, comprising the Weevil Damage Optimization Algorithm (WDOA) and the Crystal Structure Algorithm (CSA). These integrated approaches significantly elevate model precision and facilitate process optimization. Among the developed models, the CSA-enhanced version (NBCS) delivered the maximum performance and achieved a training R 2 of 0.981 for CH 4 and 0.976 for C 2 H n . More importantly, test R 2 values of 0.978 for CH 4 and 0.955 for C 2 H n confirmed the NBCS model’s strong generalization capability, outperforming the baseline NB model, which yielded test R 2 values of 0.949 and 0.955, respectively. In terms of statistical metrics, including MSE and RMSE, both NBWD and NBCS models outlined substantial improvements over the standalone NB model in predicting gasification outputs.

This study investigated the enhancement of heat transfer in turbulent air flow within a horizontal circular tube containing helical obstacles by conducting a detailed numerical analysis. Various parameters were explored, including number of helical obstacles; 1, 2, 3 and 4, helix ratio ( y  = H/D); 2, 3 and 4. Air was selected as the working fluid with Reynolds number range of 10,000 ≤ Re ≤ 60,000 to assess performance under varying flow velocities under constant heat flux along the tube surface. The cases were designed and solved using ANSYS Fluent, and k- ω turbulence model was used. The achieving results showed increasing in Nusselt number 82 % over that of plain tube. However, this was accompanied by a content raise in the friction factor. Despite this increased flow resistance, all the obtained thermal performance factors from the analyzed scenarios were greater than unity, with the highest being 1.56 achieved at n and y of 4 and 2 respectively. The configuration of four helical obstacles, n  = 4, and helix ratio y  = 2 showed the highest thermal enhancement at Re = 10,000. These results indicated that multiple helical obstacles, especially in configurations with lower helix ratios, significantly boosted heat transfer efficiency.