Hybrid LMANN Framework for Predicting Bioconvective Buongiorno Casson Nanofluid Flow over a Riga Plate
This paper develops and validates a hybrid Levenberg–Marquardt artificial neural network (LMANN) framework that accurately predicts the complex bioconvective Buongiorno Casson nanofluid flow over a Riga plate by leveraging numerical solutions from a MATLAB bvp5c solver as training data.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the world of engineering and manufacturing, fluids are rarely as simple as water. Many materials used in industry, from thick lubricants and drilling muds to biological suspensions and molten polymers, do not flow with a constant ease. Instead, their resistance to movement changes depending on how hard they are pushed or pulled. Scientists call these non-Newtonian fluids, and understanding how they move is critical for everything from coating surfaces to cooling electronic components. To make matters more complex, engineers often mix tiny solid particles, known as nanoparticles, into these fluids to boost their ability to carry heat. When these mixtures are also subjected to magnetic fields or contain living, swimming microorganisms, the physics becomes a tangled web of interacting forces. Predicting exactly how such a fluid will behave under these conditions is difficult because the equations governing the flow are highly nonlinear, meaning small changes in one factor can lead to unpredictable shifts in the whole system.
A team of researchers has tackled this challenge by developing a new way to predict the behavior of a specific, complex fluid mixture flowing over a special type of surface. They focused on a fluid that combines the properties of a Casson fluid, which behaves like a thick paste until a certain force is applied, with a Williamson fluid, which exhibits elastic properties. This mixture flows over a Riga plate, a surface equipped with electrodes and magnets that generate a specific electromagnetic force to control the fluid's motion. The study also accounts for the presence of motile microorganisms, which swim through the fluid and create their own currents, and the effects of activation energy, which influences how chemical reactions proceed within the flow. To solve this, the researchers first used traditional numerical methods to calculate the fluid's velocity, temperature, and concentration of particles and microbes. They then used these calculations to train a hybrid artificial intelligence model, specifically a Levenberg–Marquardt artificial neural network. This computer system learned the patterns in the data, allowing it to predict the fluid's behavior with high speed and accuracy without needing to solve the complex equations from scratch every time.
The researchers found that the electromagnetic force generated by the Riga plate acts as a powerful brake on the fluid's motion, effectively controlling the boundary layer where the fluid moves fastest. However, the fluid's own internal resistance plays a major role as well. Increasing the parameters that define the fluid's thickness and elasticity, such as the Casson and Williamson parameters, significantly slows down the flow. Conversely, when the buoyancy forces caused by temperature differences and the presence of microorganisms are strengthened, the fluid speeds up. The study also revealed how heat and mass are transported within this system. Factors like thermal radiation and the random movement of nanoparticles tend to raise the fluid's temperature, while the suction of fluid through the surface or a higher Prandtl number, which relates to how easily heat diffuses, cools it down. The presence of living microorganisms adds another layer of complexity; their movement is influenced by the flow speed and the concentration of other particles, creating a dynamic balance between biological activity and physical transport.
To ensure their new prediction tool was reliable, the team subjected it to rigorous testing. They compared the neural network's predictions against the detailed numerical solutions they had calculated earlier. The results showed an almost perfect match, with the artificial intelligence model predicting the velocity, temperature, and concentration profiles with negligible error. The model successfully learned the intricate relationships between the various physical parameters and the resulting flow behavior. For instance, when the researchers increased the activation energy, which represents the energy barrier for chemical reactions, the model correctly predicted that the concentration of nanoparticles and the density of microorganisms would rise because the chemical reactions consuming them slowed down. Similarly, the model accurately captured how increasing the thermal relaxation time, which accounts for the delay in heat propagation, would reduce the temperature gradient at the surface.
The study confirms that this hybrid approach, combining traditional numerical simulation with machine learning, offers a robust and efficient way to model these complex bioconvective flows. The neural network converged quickly, learning the necessary patterns in a matter of seconds and demonstrating that it could generalize well to new data without overfitting. This means the model is not just memorizing the specific cases it was trained on but has truly understood the underlying physics. The researchers conclude that this framework provides a powerful computational tool for analyzing similar nonlinear transport problems. By accurately predicting how these sophisticated fluids behave, engineers can better design systems for thermal management, chemical processing, and biomedical applications where controlling the flow of complex, particle-laden fluids is essential. The work bridges the gap between heavy mathematical modeling and rapid, data-driven prediction, offering a path forward for handling the intricate physics of modern industrial fluids.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.