Obstacle Configurations Research Articles

A scalable convolutional neural network approach to fluid flow prediction in complex environments

We evaluate the capability of convolutional neural networks (CNNs) to predict a velocity field as it relates to fluid flow around various arrangements of obstacles within a two-dimensional, rectangular channel. We base our network architecture on a gated residual U-Net template and train it on velocity fields generated from computational fluid dynamics (CFD) simulations. We then assess the extent to which our model can accurately and efficiently predict steady flows in terms of velocity fields associated with inlet speeds and obstacle configurations not included in our training set. Real-world applications often require fluid-flow predictions in larger and more complex domains that contain more obstacles than used in model training. To address this problem, we propose a method that decomposes a domain into subdomains for which our model can individually and accurately predict the fluid flow, after which we apply smoothness and continuity constraints to reconstruct velocity fields across the whole of the original domain. This piecewise, semicontinuous approach is computationally more efficient than the alternative, which involves generation of CFD datasets required to retrain the model on larger and more spatially complex domains. We introduce a local orientational vector field entropy (LOVE) metric, which quantifies a decorrelation scale for velocity fields in geometric domains with one or more obstacles, and use it to devise a strategy for decomposing complex domains into weakly interacting subsets suitable for application of our modeling approach. We end with an assessment of error propagation across modeled domains of increasing size.

Machine learning and metaheuristics in microfluidic transport characterization and optimization:CFDand experimental study integrated with predictive modelling

AbstractThis study presents a comprehensive numerical and experimental analysis on microfluidic cell lysis through computational fluid dynamics (CFD), data‐driven modelling, and multi‐objective optimization. The proposed intelligent framework integrates artificial intelligence and CFD for data generation and extraction, alongside machine learning analysis and experimental studies for transport phenomena characterization in the cell lysis process. The framework explores compound effects of various inflow Reynolds numbers and geometrical parameters, including obstacle configurations and microchannel thickness. It shows substantial effects on flow patterns and mixing in varied microfluidic designs. A surrogate model, developed via central composite design, exhibits high accuracy in assessing system functionality (). The height of the implemented baffles from its lower value to the upper bound resulted in more than 42% and 14% increase in the mixing index at low and high Reynolds numbers, respectively, with minimal impact on pressure drop. The framework introduces data‐driven modelling coupled with multi‐objective optimization by desirability function (DF), non‐dominated sorting genetic algorithm (NSGA‐II), and differential evolution (DE). In the optimization of microfluidic processes, machine learning algorithms outperform desirability‐based methods, and the DE algorithm surpasses the NSGA‐II. An optimum micromixing reducing the mixing length by over 50% and mixing index above 97% achieved, fabricated, and experimental investigations conducted to validate numerical process. Through the precise control of microfluidic variables and the exploitation of microtransfer phenomena, it is possible to enhance the efficiency and selectivity of cell lysis. This not only improves the accuracy of diagnostic information but also opens up new avenues for personalized medicine and therapeutic development.

Heat transfer and flow behavior in solar thermal collector equipped with obstacles

Solar energy, an inherently pure and abundant source of power, is integral to a myriad of daily activities. Among the various solar thermal systems, Solar Air Heaters (SAHs) are often esteemed for their economic viability and practical utility in heating spaces and drying the products.. However, sometimes the conventional SAHs are criticized for their poor performance and capability in heat transfer, making them not well organized in application. The introduction of a new method of improvement in the efficiency of heat transfer in SAHs using carefully designed obstacles with sharp corners is presented in this paper. The obstacles are conceptualized in a Triangular winglet configuration, type-based on the following: Type 1 (Single Triangular winglet), Type 2 (Dual Triangular winglet), Type 3 (Triple Triangular winglet), and Type 4 (Quadruple Triangular winglet). The setups' empirical configurations were processed for the SAH duct where the experiments were executed with a Reynolds number (Re) up to 2000 to 16,000. The obstacle parameters such as the relative transverse pitch (Pt/b), relative obstacle height (e/H), and the relative longitudinal pitch ratio (Pl/e) were clearly maintained at 4, 0.5, and 4, respectively. From the experiments, the Nusselt number was found to be significantly enhanced, up to 3.85 for the Type 3 obstacle configuration. The Thermohydraulic Performance Parameter (η) also showed a strong value for all obstacle configurations that eventually reaches a peak value of 2.29, hence presenting a comprehensive performance for heat transfer efficiency and hydraulic performance within the SAH system.

Wave trapping by freely floating obstacles in a discretely stratified fluid

In the context of the linear water wave theory, we model the interaction of time‐harmonic water waves with an infinite array of three‐dimensional freely floating bodies in a multilayer fluid through a spectral boundary value problem, consisting of a differential equation coupled with an algebraic system. We present a variational and operator formulation for the problem and derive sufficient conditions for the existence of trapped waves. We give examples of floating obstacles supporting trapped waves in a discretely stratified fluid in different obstacle configurations with a variable number of layers. Some conclusions regarding the dependence of the sufficient condition on the number of fluid layers are drawn, considering whether the sufficient condition is of the potential or algebraic type.

Trapped modes along periodic structures submerged in a two‐layer fluid with background steady flow

The trapping of linear water waves by infinite arrays of three‐dimensional fixed periodic structures in a two‐layer fluid, where the layers are considered semi‐infinite in depth, have a common interface and move each with an independent uniform velocity with respect to the ground, is studied. The existence of real solutions to the dispersion relation demands a further stability condition on the layer velocities. From the variational formulation, after certain choices of background steady flow, results a nonlinear spectral problem, which upon a sensible linearization gives a geometric condition ensuring the existence of trapped modes (within the limits set by the stability condition). Symmetries reduce the global problem to the first quadrant of the velocity space. Examples are shown of configurations of obstacles that are both independent of the layer velocities and dependent only on their difference. Future developments are suggested.

Modeling the behavior of a mobile robot using genetic algorithms in harsh ecological environment

The article is devoted to the analysis of the behavior of a mobile robot using finite state machine algorithms in order to find a path to a goal and avoid obstacles. After justifying the use of such methods, a standard deterministic finite state machine is analyzed. Next, a theory of Markov chain is applied to this algorithm, as a result of which the finite state machine becomes part of the hidden Markov model. This makes it possible to apply probabilistic methods to planning the robot's behavior. This probabilistic nature of behavior is most promising in harsh ecological environment with unpredictable configurations of obstacles. To compare the efficiency of deterministic and probabilistic finite state machines, we applied a genetic algorithm. In the numerical experiment that we conducted in the Scilab environment, we considered two main types of environments in which a mobile robot can move, in harsh ecological environment, in particular. For each type of environment, we alternately applied each of the specified behavior algorithms. For the genetic algorithm, we used one hundred individuals, which were trained over 1000 generations to find the most optimal path in the complicated environments. As a result, it was found that the deterministic finite state machine algorithm is most promising for movement in the harsh ecological environment, and the probabilistic finite state machine algorithm gives the best result in a complex environment.

Моделирование поведения мобильных роботов с использованием генетических алгоритмов

The article is devoted to the analysis of the behavior of a mobile robot using finite state machine algorithms in order to find a way to the goal and avoid obstacles. After justifying the use of such methods, the analysis of a standard deterministic finite automaton is done. Further, the theory of Markov processes is applied to this algorithm, as a result of which the state machine becomes part of the hidden Markov model. This allows you to apply probabilistic methods to modeling the behavior of the robot. This probabilistic behavior is most promising in complex environments with unpredictable obstacle configurations. To compare the efficiency of deterministic and probabilistic finite state machine, we applied a genetic algorithm. In the numerical experiment that we conducted in the Scilab software, we considered two main types of environments in which a mobile robot can move - an office-type environment and a polygonal-type environment. For each type of environment, we alternately applied each of the indicated behavior algorithms. For the genetic algorithm, we used one hundred individuals who were trained over 1000 generations to find the most optimal path in the specified environments. As a result, it was found that the deterministic finite state machine algorithm is the most promising for movement in an office-type environment, and the probabilistic finite state machine algorithm gives the best result in a complex polygonal environment.

Division-merge based inverse kinematics for multi-DOFs humanoid robots in unstructured environments

A Two-step Division-Merge (TDM) based inverse kinematics (IK) mechanism is proposed for robots with multiple degrees of freedom (multi-DOFs) in unstructured environments. For any robot with n-DOFs (n>6), a division point is selected first to divide it into a lower part subsystem from its base to the division point, together with an upper part one from the point to its end-effector. Once a collision-free configuration is found for the upper part subsystem to match with the target, the lower part subsystem is then planned to merge with the upper part subsystem at the division point. Specifically, the division point is selected such that the lower part subsystem has 6-DOFs, while the upper part subsystem has (n-6)-DOFs. Furthermore, a Reverse & Random Rotation (RRR) algorithm is also proposed to solve the IK of the upper part subsystem in unstructured environments, while the IK of the lower part subsystem is solved analytically. The proposed TDM based mechanism is verified via experiments using a lab-customized humanoid apple harvesting robot with 11- and 13-DOFs, respectively. Under the various scenarios with different target and obstacle configurations, errors in position coordinates and orientation angles between the target pose and those of the end-effector obtained with forward kinematics (FK) are calculated. Results show that the errors in position and orientation are less than 0.005 mm and 0.007°, respectively, and the average computational time is less than 2 s for the IK solution of the entire robot. Such results indicate that the proposed mechanism is reliable and capable of reducing the complexity of IK for multi-DOFs humanoid robots. With the proposed algorithm, the apple harvesting success rate of a lab-customized 11-DOFs robot reaches 89.2%, which can be further increased up to 93.3% with error compensation, while the average time starting from the robot action to the end-effector reaching the target pose of the apple is around 12.3 s.

Vapor cloud explosions in various types of confined environments: CFD analysis and model validation

In the current work, three different types of vapor cloud explosion experiments are simulated. The purpose of the simulations is twofold: firstly, to evaluate a recently developed CFD model and secondly to analyze the involved phenomena with the help of the simulation results. The proposed model, which has been implemented in the ADREA-HF CFD code, utilizes the RANS method using the Kato and Launder modification of k-e model. Combustion is modelled by taking into account the main mechanisms that contribute to the phenomenon such as chemistry, turbulence generated from the obstacles in front of the flame front, flame instabilities and turbulence generated by the flame-front itself. The CFD model is evaluated against different types of explosions in different geometries and with various fuels. Uniform premixed fuel-air mixture is considered in all cases. A large scale vented deflagration experiment in a 10 m length enclosure is firstly simulated using methane as fuel. The external explosion effect is apparent in this case. Then, a hydrogen deflagration experiment in a 78.5 m tunnel is simulated. Four mock-up cars are placed in the premixed region. Finally, propane and methane explosions inside a 1.5 m tube with obstacles and intense turbulence are simulated. Two different obstacle configurations are studied. The model predicts the overpressure values satisfactorily in all the examined cases. The factors that contribute to the pressure rise in each stage of each experiment are discussed based on the simulation results.

Multi relaxation time Lattice Boltzmann analysis of lid-driven rectangular cavity subject to various obstacle configurations

We have simulated two-dimensional laminar heat transfer in a lid-driven rectangular enclosure encapsulating heated obstacle filled with a regular viscous fluid. A square and rectangular-shaped heated blockage with different eccentric configurations is used in the physical domain to study mass and heat transfer in flow and temperature fields. We have applied D2Q9 and D2Q5 models of the Multi-Relaxation Time Lattice Boltzmann method to compute and simulate our results. The effect of dimensionless parameters like Grashof number (Gr=103, 104, 105, 106), Richardson number (Ri=0.01, 0.1, 1, 5, 10, 50, 100), Prandtl number (Pr=0.71, 1, 7) and of different eccentric placements of heated obstacle is studied on mixed convection heat transfer. The assigned value of the Reynolds number is 1000. We have also studied the impact of different aspect ratios (AR=2, 0.5, 5, 0.2) of the enclosure and the sizes of heated obstacles on fluid flow. We have also computed the local Nusselt number (Nulocal) average Nusselt number (Nu¯) along the cold wall of the enclosure.

Atmospheric Wind Field Modelling with OpenFOAM for Near-Ground Gas Dispersion

CFD simulations of near-ground gas dispersion depend significantly on the accuracy of the wind field. When simulating wind fields with conventional RANS turbulence models, the velocity and turbulence profiles specified as inlet boundary conditions change rapidly in the approach flow region. As a result, when hazardous materials are released, the extent of hazardous areas is calculated based on an approach flow that differs significantly from the boundary conditions defined. To solve this problem, a turbulence model with consistent boundary conditions was developed to ensure a horizontally homogeneous approach flow. Instead of the logarithmic vertical velocity profile, a power law is used to overcome the problem that with the logarithmic profile, negative velocities would be calculated for heights within the roughness length. With this, the problem that the distance of the wall-adjacent cell midpoint has to be higher than the roughness length is solved, so that a high grid resolution can be ensured even in the near-ground region which is required to simulate gas dispersion. The evaluation of the developed CFD model using the German guideline VDI 3783/9 and wind tunnel experiments with realistic obstacle configurations showed a good agreement between the calculated and the measured values and the ability to achieve a horizontally homogenous approach flow.

Modelling denitrification process in a static mixer–reactor using lattice-Boltzmann method

Nitrate removal in wastewater systems is crucially important to maintaining the quality of potable water as well as protecting vulnerable ecosystems. In this study, a denitrification model was incorporated into lattice-Boltzmann model (LBM) to investigate denitrification efficacy of wastewater treatment in a static mixer–reactor. The model was used for simulating the effects of six solid and porous obstacle configurations, different inlet velocities, and inlet methanol concentrations on denitrification within a static mixer–reactor with six different configurations. The results show that all the obstacles and porous blocks could improve the denitrification efficiency. However, the performances of these configurations could be affected differently by the inlet conditions. At a low velocity (u=0.5u0), the denitrification rates are almost twice as that at u=1.5u0. However, at a high inlet velocity, there were small differences of the denitrification efficiency among different configurations. Furthermore, the rectangular obstacle had the greatest effects on the denitrification rates at the low inlet velocity while its performance at a higher velocity was almost comparable to those of the porous blocks. It is also found that at a high inlet methanol concentration (C2-u = 2.0C2-u0), all the configurations increased the denitrification efficiency of up to 10%, compared to clear channel. However, the rectangular obstacle resulted in a decrease of about 3% in denitrification at a low inlet methanol concentration (C2-u = 0.5C2-u0) while all other configurations could improve the performance. Although at a low methanol concentration, the solid obstacles could dwindle the performance, the porous blocks did enhance denitrification rates in all the inlet methanol concentrations. This implies that the porous configurations have a wider range of operational conditions, compared to other obstacle configurations. A proper design of static mixer–reactors without any moving part is able to improve the denitrification efficiency in wastewater treatment. The present model is a useful tool to evaluate the performance of different reactor configurations under combinations of all significant parameters, such as inlet velocity and methanol concentrations.

An LES-DFSD study of transient premixed propane/air flames propagating past obstacles

Simulation of deflagrations requires accurate modelling of transient premixed flames. An initial laminar flame kernel is subjected to progressive stretch by obstacle generated turbulence as it propagates downstream. Large eddy simulation (LES) based flame surface density models may encounter difficulties when a predefined model constant is used for given turbulence characteristics. The reliable description of flame wrinkling considerably determines the transient flame behaviour. In this work, we applied a dynamic flame surface density (DFSD) model to automatically adjust the model parameter based on the instantaneous resolved flame information, whilst recovering laminar flame propagation in the absence of sub-grid turbulence. In the work we have successfully verified the LES-DFSD approach against recent experimental data of stoichiometric propane/air deflagrating flames past obstacles. The overpressure, flame front speed and other flame features have been correctly reproduced by the model at various stages of flame propagation for distinct obstacle configurations. The wrinkling factor was identified as an informative parameter to understand the flame dynamics and pressure build-up mechanisms, and the significant contribution from the sub-grid flame surfaces have been demonstrated. By constructing the combustion regime diagram for LES, we have demonstrated that the flame wrinkling is fully resolved during the early propagation, and the flame lies in the thin reaction zone regime when the pressure peak is reached. Through comprehensive parametric studies, we found that ignition modelling has a considerable impact on the time taken to reach the peak pressure yet a much weaker effect on the peak magnitude. A negative correlation was identified between the Smagorinsky constant and the maximum pressure. Early-stage overpressure and flame evolutions were confirmed to be grid-independent, while the pressure peak is slightly mesh-sensitive. It was found that the magnitude of the dynamic parameter adequately self-adjusts to varying grid and filter widths. The satisfactory agreements with experiments and the robust findings ensured by the sensitivity studies indicate the predictive capabilities and the benefits of the LES-DFSD model.

Explosion dynamics of hydrogen-air mixtures in a flat vessel filled with annular obstacles

Experiments were carried out in a flat vessel (205 mm inner-diameter and 70 mm height) filled with annular obstacles to study the explosion dynamics of hydrogen-air mixtures over a wide equivalence range (0.6⩽φ⩽2.4) at 0.1 MPa and 298 K. Two annular obstacles with the diameters of 100 mm (C1) and 150 mm (C2) were employed. Four obstacle configurations were used, i.e., no obstacle, a single obstacle (C1 or C2) and two obstacles (C1 and C2). Along the radius, three pressure transducers were mounted on the top wall to determine the explosion parameters at different locations. The distance between the transducers and the lid center were 0 mm (P1), 50 mm (P2) and 100 mm (P3). The maximum explosion pressure (pmax) and the maximum pressure rise rate ((dp/dt)max) in the flat vessel was found to be dependent of the measuring point. For a given equivalence ratio, the maximum explosion pressure obtained by P2 was the highest. The maximum pressure rise rate determined by P2 was the highest without an obstacle. For the cases with obstacles, (dp/dt)maxdetermined by P3 was the highest. The explosion time, combustion duration and fast explosion duration were independent of the measuring location. For a fixed measuring point, pmax was decreased by introducing an obstacle due to the extra heat loss to the obstacle. However, depending on the obstacle location, the obstacle could play a dual role in affecting the maximum pressure rise rate and time-parameters, i.e., heat sink or turbulence-generator.

An Integrated Localization, Motion Planning and Obstacle Avoidance Algorithm in Belief Space

As robots are being increasingly used in close proximity to humans and objects, it is imperative that robots operate safely and efficiently under real-world conditions. Yet, the environment is seldom known perfectly. Noisy sensors and actuation errors compound to the errors introduced while estimating features of the environment. We present a novel approach (1) to incorporate these uncertainties for robot state estimation and (2) to compute the probability of collision pertaining to the estimated robot configurations. The expression for collision probability is obtained as an infinite series and we prove its convergence. An upper bound for the truncation error is also derived and the number of terms required is demonstrated by analyzing the convergence for different robot and obstacle configurations. We evaluate our approach using two simulation domains which use a roadmap-based strategy to synthesize trajectories that satisfy collision probability bounds.

Numerical simulation of mixed convection flow and heat transfer in the lid-driven triangular cavity with different obstacle configurations

There are several industrial applications, particularly lid-driven walls, for mixed convection heat transfer characteristics across various cavities. In order to increase the effectiveness of cooling, electrical, electronic and nuclear devices and to monitor the fluid flow and heat exchange of solar thermal installations and thermal storage, such a problem requires further investigation. The main goal of this profound study is to examine the convective heat transfer nature of thermal convection on Newtonian MHD fluid in a lid-driven triangular cavity subjected to heating by a thick triangular wall, including the effects of varying Richardson number, Reynolds number, Hartmann number, and cold circular obstacle. Graphical illustration shows that the upper wall having temperature Th is moving from left to right, whereas inclined sidewalls are adiabatic. Further, a cold circular obstacle containing temperature T∗∗ is placed near the left and right wall of the triangular cavity with T∗∗ < Th. The governing flow equations are tackled by the Galerkin Finite Element Method (GFEM) to prepare the desired results. Velocity contours, isotherms and heat transfer rates demonstrate flow kinematics and heat transport processes. The high estimate of the Reynolds number has been found to provide a decent rate of convective heat transfer to a good liquid progression. For a higher number of Grashof, which achieves the maximum convection intensity, it is checked that natural convection dominates within the convection system. The Richardson number is a rising function of the Nusselt number, while the opposite trend is observed due to the rise in the Hartmann number.

Deep Reinforcement Learning Controller for 3D Path Following and Collision Avoidance by Autonomous Underwater Vehicles

Control theory provides engineers with a multitude of tools to design controllers that manipulate the closed-loop behavior and stability of dynamical systems. These methods rely heavily on insights into the mathematical model governing the physical system. However, in complex systems, such as autonomous underwater vehicles performing the dual objective of path following and collision avoidance, decision making becomes nontrivial. We propose a solution using state-of-the-art Deep Reinforcement Learning (DRL) techniques to develop autonomous agents capable of achieving this hybrid objective without having a priori knowledge about the goal or the environment. Our results demonstrate the viability of DRL in path following and avoiding collisions towards achieving human-level decision making in autonomous vehicle systems within extreme obstacle configurations.

Distribution of the random walk conditioned on survival among quenched Bernoulli obstacles

Place an obstacle with probability $1-p$ independently at each vertex of $\mathbb Z^d$ and consider a simple symmetric random walk that is killed upon hitting one of the obstacles. For $d \geq 2$ and $p$ strictly above the critical threshold for site percolation, we condition on the environment such that the origin is contained in an infinite connected component free of obstacles. It has previously been shown that with high probability, the random walk conditioned on survival up to time $n$ will be localized in a ball of volume asymptotically $d\log_{1/p}n$. In this work, we prove that this ball is free of obstacles, and we derive the limiting one-time distributions of the random walk conditioned on survival. Our proof is based on obstacle modifications and estimates on how such modifications affect the probability of the obstacle configurations as well as their associated Dirichlet eigenvalues, which is of independent interest.

Predicting porosity, permeability, and tortuosity of porous media from images by deep learning

Convolutional neural networks (CNN) are utilized to encode the relation between initial configurations of obstacles and three fundamental quantities in porous media: porosity (varphi), permeability (k), and tortuosity (T). The two-dimensional systems with obstacles are considered. The fluid flow through a porous medium is simulated with the lattice Boltzmann method. The analysis has been performed for the systems with varphi in (0.37,0.99) which covers five orders of magnitude a span for permeability k in (0.78, 2.1times 10^5) and tortuosity T in (1.03,2.74). It is shown that the CNNs can be used to predict the porosity, permeability, and tortuosity with good accuracy. With the usage of the CNN models, the relation between T and varphi has been obtained and compared with the empirical estimate.

On the Strain Hardening Behavior of a Twinning Deficient Fe–26mn–1al–0.14c Steel

The uniaxial tensile strain hardening behavior of an Fe–26Mn–1Al–0.14C steel with a medium grain size of ~ 5 µm has been investigated using X-ray diffraction and transmission electron microscopy and compared to that in the same steel treated to give a grain size of ~ 35 µm. The steel with the medium grain size exhibited no significant deformation twinning, until failure at ~ 50% true strain, yet it showed a stable high strain hardening rate i.e . ~ 2 GPa, comparable to that of the steel with the coarser grain size. The twin deficiency is explained by dearth of dislocation pile-ups and stacking faults. Grain refinement and interacting dislocations in the medium grain size steel increased its effective stacking fault energy, as determined using X-ray diffraction: it was ~ 60 mJ/m 2 compared to ~ 21 mJ/m 2 in the coarse grained steel. The dominant deformation mechanisms under these twinning deficient conditions are the formation of Taylor lattices at small strains (~ 2%) and dislocation cross-slip. The observed flow stress of the steel beyond 2% strain was interpreted as being due to lattice friction, dislocation loops and newly created mobile dislocations obeying Taylor’s hardening model. The contribution of dislocation loops to flow stress was estimated to be only ~ 33 MPa and assumed to remain constant throughout the deformation strain range. The nearly stable work hardening observed until higher strains could be explained through the cessation of cross-slip at high strains and the newly created mobile dislocations breaking away from the pre-existing dislocation obstacle configurations. The implications of this novel strain hardening mechanism based on the cross-slip assisted formation of organized dislocation patterns proposed by Kubin and Devincre is discussed.