Chinh N.Q.,Eotvos Lorand University |
Valiev R.Z.,Ufa State Aviation Technical University |
Sauvage X.,Jean Monnet University |
Varga G.,Eotvos Lorand University |
And 6 more authors.
Advanced Engineering Materials | Year: 2014
The microstructural and mechanical properties of an ultrafine-grained (UFG) Al-Zn alloy processed by high-pressure torsion (HPT) are investigated using depth-sensing indentations, focused ion beam, scanning electron microscopy and scanning transmission electron microscopy. Emphasis is placed on the microstructure and the effects of grain boundaries at room temperature. The experiments show the formation of Zn-rich layers at the Al/Al grain boundaries that enhance the role of grain boundary sliding leading to unique plastic behavior in this UFG material. The occurrence of significant grain boundary sliding at room temperature is demonstrated by deforming micro-pillars. Our results illustrate a potential for using UFG materials as advanced functional materials in electronic micro-devices. High-pressure torsion is used to produce an ultrafine-grained structure in an Al-30% Zn alloy and the mechanical behavior is investigated using various techniques. We show that Zn-rich layers form at the Al/Al grain boundaries, which enhance the role of grain boundary sliding and produce stable deformation, even at the micro-scale, without the strain avalanches, which characterize micro-sized coarse-grained crystals. © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
Corlett A.E.,UK National Engineering Laboratory
2nd World Congress on Industrial Process Tomography | Year: 2013
An experimental study of an electrical capacitance tomography (ECT) system in multiphase flow was made at the National Engineering Laboratory. Its aim was to examine the system’s flow pattern recognition and void fraction measurement capabilities. A commercial ECT system was used together with a custom-made sensor spoolpiece. The test envelope spanned a wide range of multiphase flow conditions involving mixtures of crude oil, water and nitrogen gas. The water fraction was varied between 5% and 90% and two different water phases, tap water and a MgS04 solution, were used.Results indicated that the ECT system was most suitable for flow pattern identification purposes at low water fractions, although slug and stratified regimes could be identified at higher water fractions. Void fraction measurement was not found to be particularly successful. The sensitivity map and image reconstruction algorithm used were found to be critical factors in the quality and reliability of the generated results. © 2014 International Society for Industrial Process Tomography.
MacAulay G.D.,UK National Engineering Laboratory |
Giusca C.L.,UK National Engineering Laboratory
CIRP Annals - Manufacturing Technology | Year: 2016
Structured surfaces are increasingly popular for many applications and are characterised by dimensional properties, such as diameter, often reported without an uncertainty or one based on a reproducibility analysis, which does not account for systematic effects. This paper presents a method to assess uncertainty in structured surfaces using the instrument's metrological characteristics and applies it to an example surface. Information, such as the flatness map, is used to estimate uncertainty via Monte Carlo calculations, as single values of the metrological characteristics, which are usually estimated, do not give enough information. The calculated standard uncertainty of a feature is slightly larger than that provided by a reproducibility study for the same surface, but the proposed approach accounts for additional uncertainty sources due to systematic effects. © 2016.
Li Z.,UK National Engineering Laboratory |
Song Y.,UK National Engineering Laboratory |
McLoughlin I.,University of Kent |
Dai L.,UK National Engineering Laboratory
ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings | Year: 2016
Transfer learning methods have demonstrated state-of-the-art performance on various small-scale image classification tasks. This is generally achieved by exploiting the information from an ImageNet convolution neural network (ImageNet CNN). However, the transferred CNN model is generally with high computational complexity and storage requirement. It raises the issue for real-world applications, especially for some portable devices like phones and tablets without high-performance GPUs. Several approximation methods have been proposed to reduce the complexity by reconstructing the linear or non-linear filters (responses) in convolutional layers with a series of small ones., In this paper, we present a compact CNN transfer learning method for small-scale image classification. Specifically, it can be decomposed into fine-tuning and joint learning stages. In fine-tuning stage, a high-performance target CNN is trained by transferring information from the ImageNet CNN. In joint learning stage, a compact target CNN is optimized based on ground-truth labels, jointly with the predictions of the high-performance target CNN. The experimental results on CIFAR-10 and MIT Indoor Scene demonstrate the effectiveness and efficiency of our proposed method. © 2016 IEEE.
Almahwasi A.A.,University of Surrey |
Jeynes J.C.,University of Exeter |
Merchant M.J.,University of Surrey |
Bradley D.A.,University of Surrey |
Regan P.H.,UK National Engineering Laboratory
Radiation Physics and Chemistry | Year: 2016
Ionising radiation can induce giant-nucleated cells (GCs) in the progeny of irradiated populations, as demonstrated in various cellular systems. Most in vitro studies have utilised quiescent cancerous or normal cell lines but it is not clear whether radiation-induced GCs persist in the progeny of normal replicated cells. In the current work we show persistent induction of GCs in the progeny of normal human-diploid skin fibroblasts (AG1522). These cells were originally irradiated with a single equivalent clinical dose of 0.2, 1 or 2. Gy of either X-ray or proton irradiation and maintained in an active state for various post-irradiation incubation interval times before they were replated for GC analysis. The results demonstrate that the formation of GCs in the progeny of X-ray or proton irradiated cells was increased in a dose-dependent manner when measured 7 days after irradiation and this finding is in agreement with that reported for the AG1522 cells using other radiation qualities. For the 1. Gy X-ray doses it was found that the GC yield increased continually with time up to 21 days post-irradiation. These results can act as benchmark data for such work and may have important implications for studies aimed at evaluating the efficacy of radiation therapy and in determining the risk of delayed effects particularly when applying protons. © 2016.