Yazar "Tapan, N. Alper" seçeneğine göre listele

Listeleniyor 1 - 11 / 11
  • Küçük Resim Yok
    Öğe
    Analysis and modeling of high-performance polymer electrolyte membrane electrolyzers by machine learning
    (Pergamon-Elsevier Science Ltd, 2022) Gunay, M. Erdem; Tapan, N. Alper; Akkoc, Gizem
    In this study, box and whisker and principal component analysis, as well as classification and regression tree modeling as a part of machine learning were performed on a database constructed on PEM (polymer electrolyte membrane) electrolysis with 789 data points from 30 recent publications. Box whisker plots discovered that pure Pt at the cathode surface, Ti at the anode support, the existence of Pt, Ir, Co, Ru at the anode surface, Ti porous structures at the electrodes, pure water-electrolyte and Nafion and Aquivion type membranes in proton exchange electrolyzer provide the highest performances. Principal component analysis indicated that when cathode surface consists of mostly pure Ni, when anode electrode has no support or vanadium (10-20%) doped TiO2 support and when anode electrode surface consists of cobalt-iron alloys (0.5:0.5 and 0.333:0.666 mol ratio) or RuO2, there is a risk for low-performance. Classification trees revealed that other than current density and potential, cathode surface Ni mole fraction, anode surface Co mole fraction are the most important variables for the performance of an electrolyzer. Finally, the regression tree technique successfully modeled the polarization behavior with a RMSE (root mean square error) value of 0.18. (c) 2021 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
  • Küçük Resim Yok
    Öğe
    Analysis of CO selectivity during electroreduction of CO2 in deep eutectic solvents by machine learning
    (Springer, 2023) Guenay, M. Erdem; Tapan, N. Alper
    In this work, supervised and unsupervised machine learning approaches were applied to determine routes to high CO selectivity during the electroreduction of CO2 in deep eutectic solvents (DES) utilizing the molecular, chemical, and physical characteristics of hydrogen bond donors and acceptors, as well as the properties of different electrodes and DES solvents. In addition, effective data visualization and machine learning techniques were employed to identify relationships between descriptor variables and CO faradaic efficiency. First, SHAP (Shapley Additive exPlanations) analysis was applied to determine the positive and negative effects of the descriptor variables on the target, and it was found that urea in HBD (hydrogen bond donor) has the greatest impact on the target. Then, principal component analysis (PCA) was used to identify the combinations that lead to low, medium, and high levels of the target. PCA indicated that high-level clusters may be linked with HBA (hydrogen bond acceptor) molecular properties rather than HBD in addition to choline chloride-type HBA, HBA/HBD ratio, HBD density, HBD melting point, and urea-type HBD. Finally, decision tree classification was used to discover the variables leading to very high levels of the target. The decision tree revealed one pathway with very high CO faradaic efficiency and two pathways with high CO faradaic efficiency. To conclude, future researchers will be able to design new experiments with less effort and time while analyzing the effect of new DES components for high-performance CO2 electrolyzers as a result of the machine learning study and exploratory data analysis performed in this study.
  • Küçük Resim Yok
    Öğe
    Analysis of past experimental data in literature to determine conditions for high performance in biodiesel production
    (Wiley, 2016) Tapan, N. Alper; Yildirim, Ramazan; Gunay, M. Erdem
    In this study, published experimental works on catalytic transesterification were analyzed to determine the most important variables affecting fatty acid conversion and the most suitable ranges of these variables for high performance. A database of 1324 data points was constructed from the experimental results in 31 representative papers published between 2008 and 2014, and this database was analyzed using artificial neural network (ANN) and decision tree (DT) techniques. It was found from ANN analysis that the most important variable for high fatty acid conversion was reaction time (with about 40% relative importance) followed by catalyst loading, alcohol:oil molar ratio, operating temperature, and support type with similar relative importance (about 10% each). DT analysis revealed 14 combinations of conditions leading to high performance, and some of these seemed to be generalizable for the use for the future studies; some heuristics were also derived from these generalizable conditions. (c) 2016 Society of Chemical Industry and John Wiley & Sons, Ltd
  • Küçük Resim Yok
    Öğe
    Application of Machine Learning for the Determination of Damaged Starch Ratio as an Alternative to Medcalf and Gilles Principle
    (Springer, 2023) Tapan, N. Alper; Gunay, M. Erdem; Yildirim, Nilufer
    As an alternative to the conventional amperometric method used for the determination of damaged starch ratio in wheat flour, two machine learning techniques were applied to a database constructed of 6264 voltammetric data obtained at two different electrodes, two different potassium iodide concentrations, and three different damaged starch ratios. Lift maps were extracted using association rule mining from the voltammetric database to describe electrode behavior and sensitivity to Chopin Dubois units (UCD) values. K-nearest neighbors (KNN) algorithm applied to the voltammetric experiments was able to predict UCD with higher accuracy when KI concentration was low. In addition, current quartiles, scatter, and shifts in lift maps and distinct regions after KNN classification showed that higher sensitivity towards damaged starch ratio is achieved on GC electrode and at low KI concentration.
  • Küçük Resim Yok
    Öğe
    Constructing global models from past publications to improve design and operating conditions for direct alcohol fuel cells
    (Inst Chemical Engineers, 2016) Tapan, N. Alper; Gunay, M. Erdem; Yildirim, Ramazan
    This work aims to analyze past publications on direct alcohol fuel cells (DAFC) in the literature using two data mining tools (artificial neural networks and decision trees) and to develop global models to predict the conditions leading to high performance of DAFC. The database constructed for this purpose contains 4682 data points over 271 polarization (IV) curves obtained from 36 publications in the literature. Decision tree classification models were used to develop heuristics to select the suitable fuel cell design and operational conditions to improve the maximum power density while artificial neural network models (ANN) were developed to test the predictability of IV curves at the conditions where experimental results were not available. The same ANN models were also used to determine the relative importance of design and operational variables to provide some insight to determine the variable to be manipulated. All these analyses were quite successful deducing some useful heuristics and models for the future studies from the continuously growing experience accumulated in the literature. (C) 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
  • Küçük Resim Yok
    Öğe
    Decision tree analysis for efficient CO2 utilization in electrochemical systems
    (Elsevier Sci Ltd, 2018) Gunay, M. Erdem; Turker, Lemi; Tapan, N. Alper
    In this work, a database of 471 experimental data points excerpted from 34 different publications on electro-catalytic reduction of CO2 was formed. Firstly, the database was examined by exploratory data analysis using box and whiskers plots. Then, decision tree analysis was applied to determine the significance of the variables and to reveal the conditions leading to higher faradaic efficiency, production rate and product selectivity. It was found that Cu content smaller than 71% resulted high faradaic efficiencies depending on the amount of Sn, catholyte type, applied potential and pH of electrolyte. In this case, applied potential and Cu content were found to have the highest significance among all the input variables. On the other hand, the most generalizable combination of variables leading to high level of rate occurred when the Cu content being less than 13%, using a membrane other than Selemion AMV, employing a backing layer such as TGP-H-60 and keeping the applied potential between -1.5 and -2.6 V; for which the applied potential and CO2 flow rate were determined as the highest significant variables. Finally, the most generalizable path for the case of selectivity was obtained with Sn content higher than 15% and Cu content less than 52%, which leaded to formic acid production having the highest production rates. It was then concluded that, exploratory data analysis and decision trees can provide useful information to determine the conditions leading to higher CO2-electroreduction performance that may guide the future studies in this area.
  • Küçük Resim Yok
    Öğe
    Evaluation of polymer electrolyte membrane electrolysis by explainable machine learning, optimum classification model, and active learning
    (Springer, 2023) Gunay, M. Erdem; Tapan, N. Alper
    In this work, a database of 789 experimental points extracted from 30 academic publications was used. The primary objective was to use novel machine-learning techniques to investigate how descriptor variables affect current density, power density, and polarization, and to identify rules or pathways that result in high current density, low power density, and low polarization. First, Shapley analysis was done to find and compare the magnitude of the contribution of each variable on current density as well as the positive and negative effects of all the variables. Then, correlation coefficient heat maps were provided to display the existence of any linear relationship between the input and output variables. Additionally, k-nearest neighbor classification (as an optimal model) was able to demonstrate the entire impact of all features on the outputs. Finally, the Bayesian optimization algorithm showed that the optimum performance of polymer electrolyte membrane electrolyzer could be reached with less experimental effort and time than the usual research plan. It was then concluded that machine learning methods can aid in determining the best conditions for designing a polymer electrolyte membrane electrolyzer to produce hydrogen, which can be used to guide the planning of future experiments. [GRAPHICS] .
  • Küçük Resim Yok
    Öğe
    Machine learning solutions for enhanced performance in plant-based microbial fuel cells
    (Pergamon-Elsevier Science Ltd, 2024) Gurbuz, Tugba; Gunay, M. Erdem; Tapan, N. Alper
    It is well known that numerous operational, material and design variables act upon the performance of a plantbased microbial fuel cell which is an emerging sustainable and versatile energy device like hydrogen fuel cells. However, due to the high complexity of these bioelectrochemical systems, new solutions are required to optimize performance and uncover hidden relationships between dominant fuel cell variables. For this purpose, a database of 229 observations was created for plant-based microbial fuel cells (PMFCs) with 159 descriptor variables and a target variable (maximum power density) based on experimental results from 51 recent publications. Then, some machine learning solutions like principal component analysis (PCA), classification trees and SHapley Additive exPlanations (SHAP) analysis were applied. The PCA indicated mainly two routes involving low and high chemical oxygen demand (COD) towards high maximum power density which consists of the plant family, wastewater type, support media, construction design, separator type, anode and cathode electrodes and light source. SHAP analysis revealed that the most important factors for high performance are operating temperature, natural light, soil support medium, and constructed wetland design. Finally, the classification tree successfully demonstrated nine routes towards high maximum power density which exclude the use of graphite plate cathode electrodes.
  • Küçük Resim Yok
    Öğe
    Multi-objective optimization of PEM electrolyzers using deep neural networks and gradient boost regressor-particle swarm optimization framework
    (Pergamon-Elsevier Science Ltd, 2025) Tapan, N. Alper; Gunay, M. Erdem
    In this study, a polymer electrolyte membrane (PEM) electrolyzer database with an Iridium (Ir) anode and a platinum (Pt) cathode was built using 11 descriptors (under 36 categories) with 484 observations for production of hydrogen. First, deep neural network (DNN) models were applied on the database to model four different targets: current density, power density, the product of power density and polarization as well as the ratio of current density to polarization. Then, to add some explainability to the models, the permutation feature importance analysis was applied on the trained models to find the significance of the descriptors on the targets. Following that, partial dependence plots (PDPs) were drawn to see whether the descriptors have any positive or negative effects on the targets. Potential was discovered to be the most important variable for all four targets, and a variety of anode and cathode gas diffusion layers with different membranes were found to provide optimal levels of the targets. Finally, particle swarm optimization (PSO) was used to determine optimum routes by gradient boost regressor (GBR). Optimum current density, power density, and product of power density and polarization values beyond the limits of database were extracted by GBR-PSO framework. It was also seen that holistic optimization was not possible since optimal conditions of cathode support/surface ratio and anode catalyst loading vary in a wide range for different targets.
  • Küçük Resim Yok
    Öğe
    Routes to optimum conditions of plant based microbial fuel cells by reinforcement learning
    (Pergamon-Elsevier Science Ltd, 2025) Tapan, N. Alper; Gunay, M. Erdem; Gurbuz, Tugba
    Plant-based microbial fuel cells (PMFC) are fascinating technologies that have the potential to combine plants and bacteria to produce electricity from different solid and aqueous media like constructed wetlands and wastewater treatment facilities. Although PMFCs are evolving and demonstrating promising performance results for the development of sustainable energy and water treatment, they have not reached their full potential due to issues with continuous bioenergy generation and fuel cell system optimization through plant selection, operating conditions, electrodes, and light source, all of which are critical for optimum microbial activity on the roots and exudate rhizodeposition. In light of this, the Q learning algorithm was used in this study to determine the routes that lead to the best operating and material conditions for PMFCs. A database of 231 observations from 51 recent publications with 271 descriptors (input variables) and 3 output variables (under 9 categories were used to determine the routes leading to high maximum power density, medium open circuit potential and current density. It was seen that high maximum power density routes are achievable through the nodes of stainless-steel mesh cathode with metal-based chitosan smart catalysts, bicarbonate wastewater, and anaerobic wetland sediment. Data visualizations by radar charts also exhibited similar results for cathode material and wastewater type. For medium open circuit potential, Iris pseudacorus and for medium maximum current density anaerobic sludge inoculation and steel wire mesh/nickel current collectors are found to be important indicators.
  • Küçük Resim Yok
    Öğe
    Significant parameters and technological advancements in biodiesel production systems
    (Elsevier Sci Ltd, 2019) Gunay, M. Erdem; Turker, Lemi; Tapan, N. Alper
    Biodiesel is a mixture of fatty acid esters formed by transesterification of vegetable oil, animal fat, algae oil or waste oil with an alcohol like methanol (CH3OH), ethanol (C2H5OH) or higher alcohols. There are many important catalytic variables like catalyst type and composition, support type and pretreatment conditions (i.e. calcination temperature and time) which are utilized to achieve high yields for the transesterification reaction. In addition, operational conditions such as reaction temperature, alcohol type, alcohol to oil molar ratio and stirring speed have also quite high significance. Moreover, all these variables can be optimized under supercritical conditions by novel techniques like ultrasonic and microwave irradiation or hydrodynamic cavitation. In this work, significant catalytic and operational variables for biodiesel production are reviewed. In addition, dominant parameters together with their limitations during the application of advanced technologies are investigated in detail. Then, it has been concluded that, for better control and higher yields of biodiesel production, future research works should focus on addition of co-solvents, use of longer chain alcohols, bulky structures or ionic liquids, adjustment of mode of irradiation and modification of the instrumentation or the equipment.