Earth Imagery Segmentation on Terrain Surface with Limited Training Labels: A Semi-supervised Approach based on Physics-Guided Graph Co-Training

Given earth imagery with spectral features on a terrain surface, this paper studies surface segmentation based on both explanatory features and surface topology. The problem is important in many spatial and spatiotemporal applications such as flood extent mapping in hydrology. The problem is uniquely challenging for several reasons: first, the size of earth imagery on a terrain surface is often much larger than the input of popular deep convolutional neural networks; second, there exists topological structure dependency between pixel classes on the surface, and such dependency can follow an unknown and non-linear distribution; third, there are often limited training labels. Existing methods for earth imagery segmentation often divide the imagery into patches and consider the elevation as an additional feature channel. These methods do not fully incorporate the spatial topological structural constraint within and across surface patches and thus often show poor results, especially when training labels are limited. Existing methods on semi-supervised and unsupervised learning for earth imagery often focus on learning representation without explicitly incorporating surface topology. In contrast, we propose a novel framework that explicitly models the topological skeleton of a terrain surface with a contour tree from computational topology, which is guided by the physical constraint (e.g., water flow direction on terrains). Our framework consists of two neural networks: a convolutional neural network (CNN) to learn spatial contextual features on a 2D image grid, and a graph neural network (GNN) to learn the statistical distribution of physics-guided spatial topological dependency on the contour tree. The two models are co-trained via variational EM. Evaluations on the real-world flood mapping datasets show that the proposed models outperform baseline methods in classification accuracy, especially when training labels are limited.

HEAT TRANSFER IN ROTATING, TRAILING EDGE, CONVERGING CHANNELS WITH FULL AND PARTIAL HEIGHT STRIP-FINS

A wide variety of pin-fins have been used to enhance heat transfer in internal cooling channels. However, due to their large blockage in the flow direction, they result in an undesirable high pressure drop. This experimental study aims to reduce pressure drop while increasing the heat transfer surface area by utilizing strip-fins in converging internal cooling channels. The channel is designed with a trapezoidal cross-section, converges in both transverse and longitudinal directions, and is also skewed β=120° with respect to the direction of rotation in order to model a trailing edge cooling channel. Only the leading and trailing surfaces of the channel are instrumented, and each surface is divided into eighteen isolated copper plates to measure the regionally averaged heat transfer coefficient. Utilizing pressure taps at the inlet and outlet of the channel, the pressure drop is obtained. Three staggered arrays of strip-fins are investigated: one full height configuration and two partial fin height arrangements (Sz=2mm and 1mm). In all cases, the strip fins are 2mm wide (W) and 10mm long (Lf ) in the flow direction. The fins are spaced such that Sy/Lf = 1 in the streamwise direction. However, due to the convergence, the spanwise spacing, Sx/W, was varied from 8 to 6.2 along the channel. The rotation number of the channel varied up to 0.21 by ranging the inlet Reynolds number from 10,000 to 40,000 and rotation speed from 0 to 300rpm. It is found that

Fractional model of MHD blood flow in a cylindrical tube containing magnetic particles

In recent years, the use of magnetic particles for biomedicine and clinical therapies has gained considerable attention. Unique features of magnetic particles have made it possible to apply them in medical techniques. These techniques not only provide minimal invasive diagnostic tools but also transport medicine within the cell. In recent years, MRI, drug supply to infected tissue, Hyperthermia are more enhanced by the use of magnetic particles. The present study aims to observe heat and mass transport through blood flow containing magnetic particles in a cylindrical tube. Furthermore, the magnetic field is applied vertically to blood flow direction. The Caputo time fractional derivative is used to model the problem. The obtained partial fractional derivatives are solved using Laplace transform and finite Hankel transform. Furthermore, the effect of various physical parameters of our interest has also been observed through various graphs. It has been noticed that the motion of blood and magnetic particles is decelerated when the particle mass parameter and the magnetic parameter are increased. These findings are important for medicine delivery and blood pressure regulation.

Fibre alignment and mechanical performance of carbon fibre Sheet Moulding Compounds under different preform compaction

This work investigates the effect of preform compaction on the mechanical performance and flow-induced fibre alignment of carbon fibre reinforced Sheet Moulding Compounds (SMCs). Two groups of panels have been compression moulded from reclaimed carbon fibre tows in vinyl-ester resin with low (0.5 MPa) and high (10 MPa) preform compaction pressure Additionally, a low-cost fibre orientation analysis method has been further improved in terms of reliability, and a novel flow assessment method has been developed for carbon fibre SMCs. This approach revealed greater fibre alignment with the flow direction in the lower faces of panels as a result of greater contact time with the heated mould and a lower charge viscosity at the time of pressing. As expected, greater fibre alignment in the flow direction was observed outside the initial charge coverage area in both types of panels, where the flow was greatest. Due to additional fibre flow during the high-pressure compaction stage, the mean degree of flow alignment in the high compaction panel was 47% higher than that of the low compaction panel. Improvements in the tensile stiffness (8%) and strength (32%) were also observed as a result of the high-pressure compaction stage and associated flow alignment.

How should the lift and drag forces be calculated from 2-D airfoil data for dihedral or coned wind turbine blades?

In the present work, a consistent method for calculating the lift and drag forces from the 2-D airfoil data for the dihedral or coned horizontal-axis wind turbines when using generalized lifting-line methods is described. The generalized lifting-line methods include, for example, lifting-line (LL), actuator line (AL), blade element momentum (BEM) and blade element vortex cylinder (BEVC) methods. A consistent interpretation of classic unsteady 2-D thin airfoil theory results for use in a generally moving frame of reference within a linearly varying onset velocity field reveals that it is necessary to use not only the relative flow magnitude and direction at one point along the chord line (for instance three-quarter-chord), but also the gradient of the flow direction in the chordwise direction (or, equivalently, the flow direction at the quarter-chord) to correctly determine the magnitude and direction of the resulting 2-D aerodynamic forces and moment. However, this aspect is generally overlooked and most implementations in generalized lifting-line methods use only the flow information at one calculation point per section for simplicity. This simplification will not change the performance prediction of planar rotors, but will cause an error when applied to non-planar rotors. The present work proposes a generalized method to correct the error introduced by this simplified single-point calculation method. In this work this effect is investigated using the special case, where the wind turbine blade has only dihedral and no sweep, operating at steady-state conditions with uniform inflow applied perpendicular to the rotor plane. We investigate the impact of the effect by comparing the predictions of the steady-state performance of non-planar rotors from the consistent approach with the simplified one-point approach of the LL method. The results are verified using blade geometry resolving Reynolds-averaged Navier-Stokes (RANS) simulations. The numerical investigations confirmed that the correction derived from thin airfoil theory is needed for the calculations to correctly determine the magnitude and direction of the sectional aerodynamic forces for non-planar rotors. The aerodynamic loads of upwind and downwind coned blades that are calculated using the LL method, the BEM method, the BEVC method and the AL method are compared for the simplified and the full method. Results using the full method, including different specific implementation schemes, are shown to agree significantly better with fully-resolved RANS than the often used simplified one-point approaches.

Cooling Characteristic of a Wall Jet for Suppressing Crossflow Effect under Conjugate Heat Transfer Condition

The leading edge is the critical portion for a gas turbine blade and is often insufficiently cooled due to the adverse effect of Crossflow in the cooling chamber. A novel internal cooling structure, wall jet cooling, can suppress Crossflow effect by changing the coolant flow direction. In this paper, the conjugate heat transfer and aerodynamic characteristics of blades with three different internal cooling structures, including impingement with a single row of jets, swirl cooling, and wall jet cooling, are investigated through RANS simulations. The results show that wall jet cooling combines the advantages of impingement cooling and swirl cooling, and has a 19–54% higher laterally-averaged overall cooling effectiveness than the conventional methods at different positions on the suction side. In the blade with wall jet cooling, the spent coolant at the leading edge is extracted away through the downstream channels so that the jet could accurately impinge the target surface without unnecessary mixing, and the high turbulence generated by the separation vortex enhances the heat transfer intensity. The Coriolis force induces the coolant air to adhere to the pressure side’s inner wall surface, preventing the jet from leaving the target surface. The parallel cooling channels eliminate the common Crossflow effect and make the flow distribution of the orifices more uniform. The trailing edge outlet reduces the entire cooling structure’s pressure to a low level, which means less penalty on power output and engine efficiency.

Shrinkage and Warpage in the Permanent Shape of Sape-Memory Polyurethane Parts

The shape memory effect, as the most important ability of shape memory polymers, is a working property and provides the design ability to shape memory polymer features. Shrinkage and warpage are important parameters to control the dimensional accuracy of permanent and temporary shapes of an injection moulded shape memory polyurethane (SMPU) part. In this study, the effects of injection moulding parameters on the shrinkage and warpage of the permanent shape of moulded SMPU parts were experimentally investigated. The parameters of injection pressure, melt temperature, mould temperature, packing pressure, packing time, and cooling time, were chosen as the injection moulding control factors. Taguchi�s L27 orthogonal array design table was used with six injection moulding parameters and their three levels. The results showed that the part has different shrinkage ratios in three main directions, namely, the flow direction, perpendicular to the flow direction, and the direction through the thickness. The results of the analysis of variance showed that the cooling time is the most influential parameter on both the shrinkage (except in thickness) and warpage. The shrinkage in the flow direction as well as in perpendicular to the flow direction decreased with increasing the cooling time. Warpage also decreased with increasing the cooling time. Injection pressure and melt temperature were found to be effective on shrinkage in thickness. Effects of mould temperature, packing pressure, and packing time were found to be limited. A statistically significant relationship has been noticed among shrinkage, warpage, and residual stresses during the study.

Rheo-SAXS study of shear-induced orientation and relaxation of cellulose nanocrystal and montmorillonite nanoplatelet dispersions

The shear-induced orientation in the flow direction increases with increasing shear rate. The relaxation of the CNC:MNT in comparison to CNC-only dispersion was found to be strongly retarded due to the addition of MNT, which promotes gel formation.