Biology inspired Physics

The interdisciplinary field of neuroscience has made significant progress in recent decades, providing the scientific community in general with a new level of understanding on how the brain works beyond the store-and-fire model found in traditional neural networks. Meanwhile, Machine Learning (ML) based on established models has seen a surge of interest in the High Performance Computing (HPC) community, especially through the use of high-end accelerators, such as Graphical Processing Units(GPUs), including HPC clusters of same. In our work, we are motivated to exploit these high-performance computing developments and understand the scaling challenges for new–biologically inspired–learning models on leadership-class HPC resources. These emerging models feature sparse and random connectivity profiles that map to more loosely-coupled parallel architectures with a large number of CPU cores per node. Contrasted with traditional ML codes, these methods exploit loosely-coupled sparse data ...

It is shown that the wings of bumblebees during flapping undergo pitching (feathering angle) rotation that can be characterized as a fluid-structure interaction problem. Measurements of shape, size and inertial properties of the wings of bumblebees Bombus ignitus are described that provide the necessary input data for numerical modelling. A computational fluid dynamics (CFD) solver is combined with a dynamical model that describes the time evolution of the feathering angle. An example result of the numerical simulation is shown.

Students often annotate texts while reading by highlighting, underlining, adding marks, and writing comments beside the text. These may serve various functions and will reflect each student’s goals and understanding of the text. The research team has proposed two simple biology-inspired approaches to represent the patterns of student annotations and to help hierarchical clustering method classify students based on the similarity between their annotations; the annotations produced were simple highlighting. Both approaches have instant processing speed (i.e., less than 100 ms for dealing with a batch of forty students' annotations on a 2894-word text) compared with generic hierarchical clustering algorithm implemented in Matlab and reasonable clustering accuracy (i.e., both the precision and recall values are in the range of 0.67 to 0.71) compared with the clusters created by human raters. This research further enhances the bio-inspired annotation clustering methodology by proposi...

Students often annotate texts they are reading using highlighting, underlining, and written comments and marks in the margins of the text. These may serve various functions and will reflect each student's goals and understanding of the text. This research proposes two simple biology-inspired approaches to represent the patterns of student annotations and to cluster students based on the similarity between their annotations; the annotations produced were simple highlighting. To verify the effectiveness of the proposed approaches, the research compared the processing speed of these approaches with generic hierarchical clustering algorithm implemented in Matlab and compared the accuracy of the clusters with the clusters created by human raters. The results show that both of the proposed approaches are more efficient and accurate than the generic hierarchical clustering algorithm. The proposed methodology can be implemented as an add-on to existing learning management systems and e-book readers, to automatically offer the students important notes and annotations conducted by others (either peers or students in the past) who have similar annotation behaviour pattern and style to the students.

Various propagation models have been developed to estimate the level of noise near residential areas. Predictions and measurements have proven that proper modelling of the propagation medium is of particular importance. In the present work, calculations are performed using a ray theory methodology. The ray trajectory and transport equations are derived from the linear acoustics equations for a moving medium in three dimensions. Ground and atmospheric absorption, wave refraction and diffraction and atmospheric turbulence are taken into account by introducing appropriate coefficients in the equations. In the case of a wind turbine (W/T) it is assumed that noise is produced by a point source located at the rotor centre. Given the sound power spectrum, the noise spectrum at the receiver is obtained by solving the axisymmetric propagation problem.The procedure consists of (a) finding the eigenrays, (b) calculating the energy losses along the eigenrays and (c) synthesizing the sound pressure level (SPL) by superposing the contributions of the eigenrays. In the case of a wind park the total SPL is calculated by superposing the contributions of all W/Ts. Application is made to five cases of isolated W/Ts in terrains of varying complexity. In flat or even smooth terrain the predictions agree well with the measurements. In complex terrain the predictions can be considered satisfactory, taking into account the assumption of constant wind velocity profile.Application to a wind park shows clearly the influence of the terrain on the wind velocity and consequently on the SPL.

This study addresses the aerodynamics of elastic membrane flapping wings. Several applications of the actuator disk theory to the flapping wings of insects and birds are reviewed. In previous studies, to account for spatial and temporal variance in the wake behind the flapping wings, empirical corrections were proposed for the induced velocity and power. In the present paper, a new procedure for determination of the correction factor is proposed, using membrane-type flapping-wing devices. Wind-tunnel experiments were conducted and the stroke-averaged propulsive thrust was measured on 25-cm-wingspan (flat and 9% camber) and 74-cm-wingspan flapping-wing models. Either flapping frequency or input power was held constant during the tests. Obtained thrust forces were compared to theoretical values predicted by the actuator disk theory. Empirical correction factors to the actuator disk theory were determined, providing a best fit to the experimental data when the flapping axis aligned with freestream velocity. It is noteworthy that the numerical value for the correction factor for the 25 cm cambered wing agrees with the results obtained on large insects. The theoretical corrections for angle of attack of the flapping wing give satisfactory agreements with the experimental data only for relatively low forward speeds.

Hingeless shading systems inspired by nature are increasingly the focus of architectural research. In contrast to traditional systems, these compliant mechanisms can reduce the amount of maintenanceintensive parts and can easily be adapted to irregular, doubly curved, facade geometries. Previous mechanisms rely merely on the reversible material deformation of composite structures with almost homogeneous material properties. This leads to large actuation forces and an inherent conflict between the requirements of movement and the capacity to carry external loads. To enhance the performance of such systems, current research is directed at natural mechanisms with concentrated compliance and distinct hinge zones with high load-bearing capacity. Here, we provide insights into our biological findings and the development of a deployable structure inspired by the Flexagon model of hindwings of insects in general and the hierarchical structure of the wing cuticle of the shield bug (Graphosoma lineatum). By using technical fibre-reinforced plastics in combination with an elastomer foil, natural principles have been partially transferred into a multi-layered structure with locally adapted stiffness. Initial small prototypes have been produced in a vacuum-assisted hot press and sustain this functionality. Initial theoretical studies on test surfaces outline the advantages of these bio-inspired structures as deployable external shading systems for doubly curved facades.

Students often annotate texts they are reading using highlighting, underlining, and written comments and marks in the margins of the text. These may serve various functions and will reflect each student's goals and understanding of the text. This research proposes two simple biology-inspired approaches to represent the patterns of student annotations and to cluster students based on the similarity between their annotations; the annotations produced were simple highlighting. To verify the effectiveness of the proposed approaches, the research compared the processing speed of these approaches with generic hierarchical clustering algorithm implemented in Matlab and compared the accuracy of the clusters with the clusters created by human raters. The results show that both of the proposed approaches are more efficient and accurate than the generic hierarchical clustering algorithm. The proposed methodology can be implemented as an add-on to existing learning management systems and e-book readers, to automatically offer the students important notes and annotations conducted by others (either peers or students in the past) who have similar annotation behaviour pattern and style to the students.

Students often annotate texts while reading by highlighting, underlining, adding marks, and writing comments beside the text. These may serve various functions and will reflect each student’s goals and understanding of the text. The research team has proposed two simple biology-inspired approaches to represent the patterns of student annotations and to help hierarchical clustering method classify students based on the similarity between their annotations; the annotations produced were simple highlighting. Both approaches have instant processing speed (i.e., less than 100 ms for dealing with a batch of forty students' annotations on a 2894-word text) compared with generic hierarchical clustering algorithm implemented in Matlab and reasonable clustering accuracy (i.e., both the precision and recall values are in the range of 0.67 to 0.71) compared with the clusters created by human raters. This research further enhances the bio-inspired annotation clustering methodology by proposi...

Numerical aero-acoustic analysis of wind turbine blades is conducted on three different sized machines 350kW, 2MW &3MW. The turbine blade is considered as independent noise source and sound pressure levels produced from airfoil self-noise are assessed according to the BPM model. This model is semi empirical and predicts the sound pressure levels for 1/3 rd octave band frequencies 20Hz – 10kHz range. The frequency spectra of noise mechanism components for untripped boundary layer and directivity pattern as function of observer distance is illustrated. Model sensitivity is calculated using the directivity at several octave band frequencies and observer azimuth positions. The total sound pressure levels for the turbines are compared with theoretical predictions.

In the present study, inertial forces and kinematics
of wings of locusts, Schistocerca americana, were investigated
experimentally. The developed experimental setup includes
freshly extracted hindwings and forewings, a mechanical
transmission producing pitching and flapping, a vacuum
chamber, and a high-speed video system. Flapping angle amplitudes
determined at the middle section are practically the
same in air and in vacuum in both forewings and hindwings.
Pitching amplitudes in the root and in the middle section of
hindwing differ by up to 50 % due to torsional deformation or
twist. In air, the average twist angles are 3° and 17° in forewings
and hindwings, respectively. Amplitudes of twist angle
are higher in vacuum, especially at relatively high flapping
frequencies. Inertial forces are calculated based on a rigidbody
model of a wing with prescribed displacements. The 5-
segment mesh for hindwings and 2-segment mesh for forewings
provide accurate determination of inertial forces.
Amplitudes of the fifth harmonic in accelerations in air are
substantially smaller than in vacuum and, therefore, inertial
forces in air are about 30 % smaller than those in vacuum.
Thus, aerodynamic loads suppressed high-frequency component
in wing oscillations resulting in substantially smaller inertial
forces in air.

Aerodynamic and inertial forces and corresponding kinematics of flapping wings of locusts,
Schistocerca americana, were investigated in a low-speed wind tunnel. The experimental setup
included live locusts mounted on microbalance synchronized with a high-speed video system.
Simultaneous measurements of wing kinematics and forces were carried out on three locusts at 7°
angle of attack and velocities of 0ms−1 and 4ms−1. Time variations of flapping and pitching angles
exhibit similar patterns in fore- and hindwings and among the animals. Significant tip to root
variations in pitching angle are found in both wings. The locusts have much larger flapping and
pitching amplitudes in still air causing larger oscillations in inertial forces. Inertial forces are added to
the lift and thrust on one part of the stroke, resulting in higher reaction forces and subtracted on the
other part. Plots of the lift demonstrate similar trends with and without the wind. The global maxima
and peak-to-peak amplitudes in lift are about the same in both tests. However, local minima are
significantly lower in still air, resulting in much smaller stroke-averaged lift. Amplitudes of thrust
force oscillations are much higher in still air; consequently, the stroke-averaged thrust is higher
compared to the non-zero freestream velocity case.

The aim of wind tunnel tests is the simulation of the flow around bodies or their scaled models. In aeronautical applications, the measurement of aerodynamic loads in a wind tunnel, forces and momentums, is a very difficult task due to the required accuracy. The wind tunnel balances, comprised by several hardware and software components, provides directly the pursued measurements, with high accuracy and reliability. For these reasons, among others, wind tunnel balances have become a common tool in testing facilities.