Cetin Canpolat

Introduction: In the present study, fluid and mass transport along a microchannel are aimed to be solved using physics-informed neural networks (PINN), which is an emerging solution method for partial differential equations (PDEs). For verification purposes, the same cases under identical conditions are modeled with a solver for finite element analysis (FEA), COMSOL Multiphysics. Materials and methods: Continuity, Navier-Stokes, and Nernst-Planck Equations, which characterize momentum and mass conservation, are solved simultaneously by PINN as well as COMSOL Multiphysics. The problems are defined as steady or time-dependent with varying Reynolds numbers in the interval of 5 × 10 ⁻⁶ ≤ Re ≤ 5 × 10 ⁻² . The fluid is flowed by either pressure-driven or electrokinetically-driven. PDEs are nondimensionalized by scaling the quantities for PINN solving on the purpose of alleviating the computational burden. In contrast to usual approach, where the pressure scale is viscosity multiplied by the velocity scale divided by the length scale, pressure scale is determined as its maximum value. Results and discussion: Calculation errors of PINN solver are observed as reasonably low, such as 3.04 × 10 ⁻⁵ , 4.76 × 10 ⁻³ , 3.60 × 10 ⁻⁴ , 6.02 × 10 ⁻³ for steady pressure-driven flow, time-dependent pressure-driven flow, steady electroosmotic flow, and time-dependent electroosmotic flow, respectively. To reduce the error margin at these values, pressure scale must be defined as the maximum pressure value. Therefore, the results of the FEA solver exhibit excellent overlap with the data obtained from PINN. For PINN, species concentrations of 0.52–1.75–2.71–3.36 mol/m ³ are accumulated at the outlet under pressure-driven flow for time points 50–100–150–200 seconds, respectively. Under electrokinetic flow, species concentrations are varied as 0.72–1.93–2.83–3.46 mol/m ³ for the same time points, respectively. In consideration of FEA, species concentrations are calculated as 0.52–1.75–2.71–3.39 mol/m ³ and 0.55–1.85–2.84–3.52 mol/m ³ under similar conditions for pressure-driven and electroosmotic flows, respectively. Conclusion: Both Navier-Stokes and Nernst-Planck Equations can be solved with PINN in a single model, regardless of the flow generation method and the time dependency of the model. Because nearly identical calculations are carried out by the FEA solver. As a result, PINN is a notable alternative for simultaneous modeling of the flow and mass transport under pressure-driven and electrokinetic conditions for microfluidic applications.

A significant elevation in xanthine oxidase activity in the urine samples is reported as a diagnostic indicator of urinary tract infection (UTI). In this study, electrochemical techniques, including cyclic voltammetry (CV) and amperometry, are employed to estimate numerically xanthine oxidase activity, as electrochemical methods are easily adapted to point‐of‐care tests. The design of the second generation of biosensors is adopted; thus, mediators of ferro/ferricyanide are used as electron carriers between the enzyme and the working electrode. Hence, the disadvantages of the conventional cell culture method, such as relatively high costs, accumulation of large amounts of samples in hospital laboratories, and delayed acquisition of results, are prevented. During the analysis, the initial enzyme concentrations are systematically increased from 0.005 to 0.2 mM to correlate enzyme activities with the current responses of the electrode. For comparison, the calculations are performed using physics‐informed neural network (PINN) and COMSOL Multiphysics, which is a finite element solver. It is observed that both solvers calculate consistent current responses significantly, and the resulting currents exhibit linear behavior for increasing enzyme activity. As a result, the proposed calibration equations for CV and amperometry can be utilized with confidence to quantify the xanthine oxidase activity in the urinary media.

Bu çalışmada bir mikroakışkan sisteme entegre edilmiş biyosensörün üzerinde oluşan redoks tepkimeler çevrimsel voltammetri yöntemiyle izlenerek akışkan içerisindeki glukozun tayini yapılmıştır. Mikrokanala entegre biyosensörün tasarımı ve sonlu elemanlar analizi için Comsol Multiphysics v3.5a kullanılmıştır. Bu bağlamda, yüksekliği 300 µm ve uzunluğu 1500 µm olan bir mikrokanal yüzeyine elektrot yerleştirilmiştir. Matematiksel modelleme öncelikli olarak mikrokanal boyunca sıvı akışı olmadan gerçekleştirilmiştir. Analizlere mikrokanal girişine laminer akış koşulları altında bir sıvı girişi tanımlanarak devam edilerek çevrimsel voltammetri tabanlı bir mikroakışkan sistem tasarımı yapılmıştır. Biyosensörün etkinliği 0.25mM, 0.5mM, 1mM, 2mM, 4mM ve 8mM gibi farklı glukoz değerlerinde test edilmiştir. Mikrokanalın girişine tanımlanan laminar akışın debisinin biyosensörden alınan yanıta etkisinin belirlenmesi için 25 µl/dk, 50 µl/dk, 100 µl/dk ve 200 µl/dk gibi farklı debi değerlerinde çözümlemeler yapılmıştır. Elektrot uzunluğunun çevrimsel voltammetriden alınan cevaba etkisinin incelebilmesi için elektrot uzunluğu 100µm, 200µm ve 300µm olarak değiştirilmiştir. Bu çalışmada laminer akış varlığında çevrimsel voltammetriden elde edilen maksimum akımın yükseldiği gözlemlenmiştir. Simülasyonlar, sıvı akış debisinin arttırılmasının biyosensörün etkinliğini arttırdığını göstermiştir. Sonuç olarak, bu çalışma ile çevrimsel voltammetri tabanlı bir mikroakışkan sistem ile glukoz tayininin yapılabilirliği ortaya konulmuştur.

In this work, periodic vortex shedding at both sides of a circular cylinder is aimed to be suppressed using a concentrically located perforated cylinder under laminar flow conditions at Re=200. A code is developed in Comsol Multiphysics, 5.3a and validated using the outcomes of Ansys, Fluent and previous studies in the open literature, which exhibit a good agreement. The porosity, β, and is varied within 0.5≤β≤0.9, and the gap ratio, D/d is varied within 1.5≤D/d≤3.5 to determine the optimum parameters. The results of the present numerical investigation are evaluated using instantaneous and time-averaged vorticity, streamwise and transverse components of the velocity and pressure. The drag, C_D and lift, C_L coefficients are calculated. Strouhal number, St from the pointwise spectral analysis of the streamwise velocity component is plotted for various cases. It is observed that porosity βhas a dominant effect rather than the gap ratio on the flow past a solid cylinder. The low-velocity and low-pressure regions are getting large in the transverse direction as the porosity increases. The separated layers from solid and perforated cylinders merge for low gap ratios. However, individual movement of these layers is evident for larger gap ratios with low porosity,β values. A perfect suppression of the periodicity of vortex shedding is obtained for the cases of D/d=3.5 with β=0.5,0.6 and D/d=3 with β=0.5.

This study presents the mechanism of adsorption of a model protein within an electrical double layer (EDL) of a charged silica NP in an aqueous salt medium. In contrast to previous works, the charge regulation model analyzes the surface chemistry and electrostatic charge within the EDL by considering polarization, morphology, topography, and protonation/deprotonation. This model is coupled with the three protein adsorption models, such as Classical Langmuir (CLM), Extended Langmuir (ELM), and Two-State (TSM). By this means, surface curvature, conformational changes, and size- and pH-dependent characteristics of protein adsorption are allowed to be observed together in the present numerical model. For a parametric investigation, ionic concentration and pH of the solution, surface charge density, initial protein concentration, electrostatic charge of the protein, and the diameter of the NP are varied systematically. The validation studies with the experimental data agree well with the literature. The size-dependent protein adsorption is simulated and revealed with the current numerical model. Space charge density, chemistry, and morphology of NP’s surface profoundly influence adsorption characteristics.

In the present study, the adsorption of a protein on a nanoparticle with a nanostructured surface, which is created using successively patterned Gaussian pillars (GPs), is simulated by considering the charge regulation within the electrical double layer of a silica nanoparticle (NP). Namely, the mathematical models for the adsorption mechanism, such as classical Langmuir model, extended Langmuir model, and two‐state model, are coupled with charge regulation model. By this means, size and pH variables are able to included to the calculations. Moreover, free space, surface curvature, and conformational changes are also taken into account. For systematic investigation, the solution's pH, surface charge density, initial protein concentration, electrostatic charge of the protein, and the diameter of the spherical NP are varied. As a result, the vital properties of a nanoparticle, such as protonation/deprotonation, polarization, topography, and morphology, are considered in the current simulations. The surface charge density and surface chemistry change with NP and GP sizes. The present results reveal that the protein adsorption on an NP with a smooth surface reaches a faster complete surface coverage than an NP with a nanostructured surface. Both states of conformational changes are also affected by the presence of the GP.

The electrospun nanofibrous mat is an emerging tool for drug release studies. Therefore, it is essential to predict this type of system's drug release behavior to overcome the existing problems and generate novel drug release systems. In the present work, a numerical study is performed for fulfilling this requirement, and the current numerical data is validated with the experimental study, which is available in the open literature. Transient Langmuir-Freundlich adsorption-desorption isotherm is employed for describing the drug release behavior of the electrospun nanofibrous system under perfect sink conditions. The effect of the diffusion phenomenon is also taken into account. Drug release rates are investigated for different initial drug concentrations, porosity values, permittivity mass coefficients, and mat surface areas. Moreover, the relationship between the porosity and the initial drug concentration is also presented. It can be reported that significant alterations occur in drug release rates through varied initial drug concentrations and porosity. Although the drug release rate is altered with permittivity coefficient or surface area, minor variations are observed compared to the parameters above. The results of the numerical code agree well with the experimental data.