Browsing by Author "Hanekom, M."

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Evidence that the spread of Mycobacterium tuberculosis strains with the Beijing genotype is human population dependent
    (American Society for Microbiology, 2007-07) Hanekom, M.; Van der Spuy, G. D.; Gey van Pittius, N. C.; McEvoy, C. R. E.; Ndabambi, S. L.; Victor, T. C.; Hoal, E. G.; Van Helden, Paul D.; Warren, Robin M.
    This study describes a comparative analysis of the Beijing mycobacterial interspersed repetitive unit types of Mycobacterium tuberculosis isolates from Cape Town, South Africa, and East Asia. The results show a significant association between the frequency of occurrence of strains from defined Beijing sublineages and the human population from whom they were cultured (P < 0.0001). Copyright © 2007, American Society for Microbiology. All Rights Reserved.
  • Thumbnail Image
    Item
    Inverse method for static load reconstruction with sensitivity filtering and optimal sensor placement
    (Stellenbosch : Stellenbosch University, 2021-03) Hanekom, M.; Venter, Gerhard; Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.
    ENGLISH ABSTRACT: Static load reconstruction is a technique that uses the surface strain response of a prototype structure, in conjunction with a numerical model, to determine the magnitude of the applied loads with a least-squares estimate. A unit-load finite element (FE) model describes the relationship between every input load and the elemental output strain where each finite-element represents a potential location for mounting a strain gauge. The candidate set contains every desired potential gauge location-orientation combination (or candidate point). It is impossible to mount gauges at every candidate point, but with a D-optimal design, the optimal sensor placement can be determined. The response is assumed linear in order for the superposition principle to hold. The D-optimal design selects elements that span the maximum volume of the candidate set, but makes no consideration for the practical aspects associated with mounting strain gauges. For example, during pilot studies, the design chose locations which were either inaccessible or where gauges measured incorrect strain values. A variety of filters have been designed to exclude specific elements from the candidate set to prevent any practical difficulties in mounting strain gauges. These filters are not limited to load reconstruction and can also be used in other strain gauge operations. The structural filters remove elements at open and sharp edges as well as triangular finite-elements. Next, incorrect values will be measured if the strain output is below the measuring capability of the sensor; thus, another filter excludes elements whose numerically computed strain values are below this threshold. Experience has taught that a strain gauge should be mounted in an area of high strain, but of low strain gradient. The final alter is a statistical algorithm that considers high strain gradients to be outliers and utilises an adjusted boxplot method to remove candidate points that are associated with a high strain gradient. Numerical experiments investigated the optimisation of various versions of the candidate set, as well as how the number of candidate points in the design matrix affects the reconstructed loads. All methods worked adequately, and more points in the design (proportionality) matrix improved the accuracy with which loads were calculated. During a physical experiment, it was found that a weighted average proportionality matrix should be used to reconstruct the applied loads if a strain gauge is glued over more than one finite-element. Furthermore, the most signifcant source of error between the calculated and actual loads originates from the differences between the FE and the actual model.