Deposition Geometrical Characteristics of Wire Arc Additive Manufactured AA2219 Aluminium Alloy With Cold Metal Transfer Pulse Advance Arc Mode

A cold metal transfer pulse advance (CMT-PA) arc mode was employed in this paper for the additive manufacturing of Al alloy. The effects of process parameters on the surface morphology and effective width percentage were investigated. And a deposition width model was built by the multiple linear regressions. Based on the principle that the volume of sample is equal to that of filler wire, a deposition height model was simultaneously derived. The results show that the process parameters affect the trends of droplet spreading in horizontal direction and molten pool tangential direction by changing the heat input and arc force. The disparity between two trends directly determines the final deposition geometrical characteristics. The influences of three factors on the effective width percentage show a trend of first increasing and then decreasing. So it provides a process window of good deposition forming. Using the optimal parameter in the window, the effective width percentage reaches to 83% and machining allowance is only 0.8 mm, which significantly improves materials utilization and reduces manufacturing costs. Besides, the error rates of deposition width and height models are less than 4% and 6%, respectively. Two models can facilitate manufacturing different size parts and make a profit for the actual production.

The critical influence of some “tiny” geometrical details on the stress field in a Brazilian Disc with a central notch of finite width and length

The role of some geometrical characteristics of the notches ma­chined in circular discs, in order to determine the mode-I fracture tough­ness of brittle materials, is discussed. The study is implemented both analyti­cally and numerically. For the analytic study advantage is taken of a recently intro­duced solution for the stress- and displacement-fields developed in a finite disc with a central notch of finite width and length and rounded corners. The vari­ation of the stresses along strategic loci and the deformation of the peri­me­ter of the notch obtained analytically are used for the calibration/validation of a flexible nu­mer­ical model, which is then used for a parametric investiga­tion of the role of geometrical features of the notched disc (thickness of the disc, length and width of the notch, radius of the rounded corners of the notch). It is con­cluded that the role of the width of the notch is of critical im­port­ance. Both the ana­lytic and the numerical studies indicate definitely that ignoring the ac­curate geo­metric shape of the notch leads to erroneous results concerning the actual stress field around the crown of the notch. Therefore, it is possible that misleading values of the fracture toughness of the material of the disc may be obtained.

On the Geometrical Properties of the Lightlike Rectifying Curves and the Centrodes

This paper mainly focuses on some notions of the lightlike rectifying curves and the centrodes in Minkowski 3-space. Some geometrical characteristics of the three types of lightlike curves are obtained. In addition, we obtain the conditions of the centrodes of the lightlike curves are the lightlike rectifying curves. Meanwhile, a detailed analysis between the N-type lightlike slant helices and the centrodes of lightlike curves is provided in this paper. We give the projections of the lightlike rectifying curves to the timelike planes.

Numerical investigation of fuel flexibility in a small-scale flameless combustor

Numerical simulation of a laboratory flameless combustor was performed to investigate the flexibility to burn alternative fuels to natural gas. The studied fuels are biogas, syngas and a mixture of ammonia and methane. The inlet temperatures of air and fuel, the equivalence ratio and the geometrical characteristics of the combustor were maintained constant. The results show that flameless combustion is observed in the biogas and in the NH3/CH4 mixture, while the syngas burns according to the conventional non-premixed combustion mode. According to the predictions, the biogas emits 1.1 ppm of NOx and 229 ppm of CO, syngas produces 7.8 ppm of NOx and 35 ppm of CO and the NH3/CH4 mixture emits about 3900 ppm of NOx and 608 ppm of CO. The high NOx and CO emissions in the NH3/CH4 mixture show that the combustor needs to be optimized to burn a nitrogen-containing fuel.

Ultrasounds Energy as an Agent of Polyelectrolyte Modification Prior to Sewage Sludge Conditioning

The presented research concerned the phenomenon of polyelectrolyte changes resulting from modification by applying the ultrasonic field. The main aim of this research was to determine the activation degree of this macromolecular chemical compound and its effect on sewage sludge subjected to conditioning and followed by dewatering. The overall goal was to investigate the potential way of reducing the dosage of chemical compounds prior to sewage sludge conditioning. The polyelectrolyte samples were sonicated with the ultrasonic disintegrator UD-20 coupled with a sandwich concentrator. The power output of the generator was 180 W and the ultrasonic field frequency was 22 kHz. To describe the geometrical characteristics of the separated phases, the following parameters were determined: surface area (AA), perimeter (LA) and non-dimensional coefficient. With reference to the obtained results, the most significant quantitative changes in shape and size of the separated phases were observed for the ultrasonic field exposure time in the range of 0 to 10 s. This was in agreement with the results observed during dewatering of the investigated sewage sludge. In view of the quantitative analysis of the structure of the polyelectrolyte subjected to the ultrasonic modification, dewatering of sewage sludge was considerably improved by the application of the presented method.

Positronium Confined in Nanocavities: The Role of Electron Exchange Correlations

Positronium atoms (Ps) are commonly employed as a probe to characterize nanometric or subnanometric voids or vacancies in nonmetallic materials, where Ps can end up confined. The annihilation lifetime of a trapped Ps is strongly modified by pickoff and depends on the cavity size and on the electron density in the confining cavity surface. Here, we develop a theory of the Ps annihilation in nanocavities based on the fundamental role of the exchange correlations between the Ps-electron and the outer electrons, which are not usually considered but must be considered to correctly theorize the pickoff annihilation processes. We obtain an important relation connecting the two relevant annihilation rates (for the p-Psand the o-Ps) with the electron density, which has the property of being totally independent of the geometrical characteristics of the nanoporous medium. This general relation can be used to gather information on the electron density and on the average cavity radius of the confining medium, starting from the experimental data on PALS annihilation spectra. Moreover, by analyzing our results, we also highlight that a reliable interpretation of the PALS spectra can only be obtained if the rule of 1/3 between the intensities of p-Psand o-Pslifetimes can be fulfilled.

Geometric Regularization of Local Activations for Knowledge Transfer in Convolutional Neural Networks

In this work, we propose a mechanism for knowledge transfer between Convolutional Neural Networks via the geometric regularization of local features produced by the activations of convolutional layers. We formulate appropriate loss functions, driving a “student” model to adapt such that its local features exhibit similar geometrical characteristics to those of an “instructor” model, at corresponding layers. The investigated functions, inspired by manifold-to-manifold distance measures, are designed to compare the neighboring information inside the feature space of the involved activations without any restrictions in the features’ dimensionality, thus enabling knowledge transfer between different architectures. Experimental evidence demonstrates that the proposed technique is effective in different settings, including knowledge-transfer to smaller models, transfer between different deep architectures and harnessing knowledge from external data, producing models with increased accuracy compared to a typical training. Furthermore, results indicate that the presented method can work synergistically with methods such as knowledge distillation, further increasing the accuracy of the trained models. Finally, experiments on training with limited data show that a combined regularization scheme can achieve the same generalization as a non-regularized training with 50% of the data in the CIFAR-10 classification task.