Shaft Cooling Investigation based on CFD
Dr James Widmer, Newcastle University
This project describes research into shaft-cooling systems for electrical machines. In such shaft-cooling systems, the liquid is made to flow through the shaft hole in order to cool the machine down. The degree of heat transfer of the liquid was measured using a CFD (computational fluid dynamics) method. The results from three different turbulence models were compared. It was then found that rotation can significantly increase the convective heat transfer in the shaft hole above the stationary condition.
Parallel Computational Topology
Dr Hamish Carr, University of Leeds
Scientific analysis and visualisation depend among other things, on topological data structures, notably the contour tree, Reeb Graph, Morse-Smale Complex & Reeb Space. While these structures are increasingly important, effective parallel algorithms have lagged behind serial algorithms, and this project aims to develop OpenMP, MPI, nVidia Thrust and hybrid algorithms to analyse massive datasets beyond the capabilities of desktop workstations.
Development of Parallel, Open Source, Fluid Structure Interaction Software
Dr Lee Margetts, University of Manchester
The project aims to improve the scalability and parallel performance of an interface between the computational fluid dynamics solver OpenFOAM and finite element solver ParaFEM. To produce open source fluid structure interaction software capable of solving large models, with applications in onshore and offshore wind energy devices. The facilities at N8 provide the resources to complete such a project.
Automated Soundscape Identification
Dr David Chesmore, University of York
The project aims to develop methods for automatically identifying sounds within an urban environment. At present noise pollution is measured simply by loudness without knowledge of the sound source. We aim to use artificial intelligence to identify sounds in addition to loudness which will help decision and policy makers. The is a sub-project of the CAPACITIE EU Research Project for environmental pollutant monitoring involving identification of sound in urban soundscapes, which is computationally highly intensive. The algorithms and techniques used require days for each training session for machine intelligence algorithms.
EvoSuite – Automated Test Generation for Java Classes
Dr Gordon Fraser, University of Sheffield
To find defects in software, one needs test cases that execute the software systematically, and oracles that assess the correctness of the observed behaviour when running these test cases. The EvoSuite project works on tools to automatically generate test cases with assertions for classes written in Java code. To achieve this, EvoSuite applies novel hybrid approaches that generate and optimise whole test suites towards satisfying coverage criteria. The EvoSuite project further investigates generations of possible oracles by adding small and effective sets of assertions that concisely summarise the current behaviour. These assertions allow the developer to detect deviations from expected behaviour, and to capture the current behaviour in order to protect against future defects.