UDK 620.178.3:669.715 Original scientific article/Izvirni znanstveni članek ISSN 1580-2949 MTAEC9, 47(3)357(2013) AUTOMATED DIAGNOSTICS OF DAMAGE TO AN ALUMINUM ALLOY UNDER THE CONDITIONS OF HIGH-CYCLE FATIGUE AVTOMATIZIRANA DIAGNOSTIKA POŠKODBE ALUMINIJEVE ZLITINE PRI VISOKO-CIKLIČNEM UTRUJANJU Pavlo Maruschak1, Igor Konovalenko1, Mykhailo Karuskevich2, Vladimir Gliha3, Tomaž Vuherer3 JTernopil Ivan Pul'uj National Technical University, 46001 Ternopil, Ukraine 2National Aviation University, Aircraft Design Department, Komarova Ave. 1, 03058 Kyiv, Ukraine 3University of Maribor, Faculty of Mechanical Engineering, Smetanova 17, 2000 Maribor, Slovenia maruschak.tu.edu@gmail.com Prejem rokopisa — received: 2012-10-02; sprejem za objavo - accepted for publication: 2012-11-12 An identification and quantitative analysis of the deformation relief of the aluminium alloy for an aircraft construction based on a digital-image processing has been performed. The behaviour of defects has been assessed on the basis of diagnostics results for individual stages of the deformation process. It has been established that the individual stages of the damage-accumulation process are characterised by the values of integral-image parameters. Based on the consecutive processing of the data on the surface cyclic deformation, the main regularities of the propagation of defects have been found. Theoretical preconditions have been substantiated and experimental results obtained. Keywords: fatigue, surface, digital image, diagnostics, accumulated damage, defect propagation, evaluation Identificirali in kvantitativno analizirali smo relief deformirane površine aluminijeve zlitine za gradnjo letal. Temelj obojega je izdelava digitalne podobe površine vzorcev. Oceno vedenja defektov omogočajo rezultati diagnostike posameznih stopenj deformacijskega procesa. Ugotovili smo, da so te stopnje med akumulacijo poškodbe značilne po vrednostih integralnih parametrov slike. Na osnovi zaporedne obdelave podatkov ciklične deformacije na površini smo ugotovili glavne zakonitosti pri širjenju defektov. Tako smo dokazali teoretične predpostavke in pridobili eksperimentalne podatke. Ključne besede: utrujanje, površina, digitalna podoba, diagnostika, akumulirana poškodba, širjenje defekta, vrednotenje 1 INTRODUCTION An analysis of the loading conditions of modern civil aircrafts, the existing methods for evaluating the accumulated fatigue damage, the peculiarities of fatigue damage of aviation structural materials and the results of the previous fatigue investigations allowed formulating an approach to solving the problem of a quantitative evaluation of the accumulated fatigue damage of the aircraft structural elements1. However, the technological complexity of many of the existing instrumental methods for evaluating the accumulated fatigue damage as well as their insufficient accuracy and reliability limit the use of these methods for practical purposes.2 The initial diagnostics of the contemporary aircraft- skin condition involves the search and identification of the fatigue damage using the visual-control methods. It is known that the incubation period of the fatigue damage accumulation is, in many cases, reflected in the visual signs, which determine the possibility of both qualitative and quantitative evaluations of the accumulated fatigue damage.3 The quantitative evaluation of the accumulated damage at the initial stage of fatigue allows predicting the place and the time of a fatigue-crack appearance. At the stage of designing aviation equipment such a prediction reduces the cost of a full-scale fatigue tests significantly due to shortening their duration, and, at the stage of operation, it allows increasing the reliability of aircrafts and safety of flights.4 A deformation relief is formed on the surface of the cladding layer of aluminium alloys under the stresses corresponding to the loading conditions of many structural elements during operation and testing. Stress concentration causes a prior formation of the relief in the vicinity of riveting holes, glue-cooking points, etc., which are the areas of potential failures. A surface deformation relief is observed at several scale levels. Using the optical microscopy the signs of a relief can be observed at the meso- and macro-levels.5 The need for substantiating and implementing the objective indicators of the deformation-relief intensity as the characteristics of the accumulated fatigue damage is obvious. The solution of this problem, by means of an analysis of damaged-surface images, will be shown below. The dependencies that allow predicting the residual life are of the most practical importance. Such dependences can be obtained on the basis of the data on deformation-relief-parameter evolution.6 It is proposed herein to use the integral parameters obtained by analysing the investigated surface images. They allow evaluating the fatigue damage of the aluminium alloy for the aircraft construction. 2 DEFORMATION-RELIEF-EVALUATION TECHNIQUE The specimen geometry is shown in Figure 1; the specimen was tested with cantilever bending at R = 0, tfmax = 147 MPa. The analysis of the deformation relief was performed near the stress concentrator (a hole with a diameter of 1.0 mm). In the process of testing we recorded the surface condition in the vicinity of the stress concentrator after an application of cyclic loading, the number of loading cycles necessary for the initiation of a fatigue crack with a length of 1.0 mm, and the number of loading cycles leading to a complete failure. The specimen surface condition was evaluated by analysing its photo images obtained in a series after a certain number of loading cycles. The images taken with a photo camera (Figures 1b to d) were transformed into grayscale images with the brightness function I(x,y). The absolute values of horizontal and vertical gradients were calculated for every pixel of an image: VI x (x, y) = dI( x, y) dx VI y (x, y) = dI (x, y) dy (1) where x e Nm, y e Nn (m and n are the width and height of an image, respectively): Figure 1: a) Scheme of the specimen investigated; examples of the deformation relief after 15, 100, 711 thousand loading cycles: b), c), d) initial image and e), f), g) binary image Slika 1: a) Shema vzorca za preiskavo; primer reliefa deformirane površine po 15, 100, 711 tisoč obremenilnih ciklih: b), c), d) začetna slika in e), f), g) binarna slika The mean values of horizontal and vertical gradients were used for a generalised evaluation of the condition of the surface investigated: Gx =VIX = — x mn Gy = VIy = — y y mn 1 n m - JJWx te y)dxdy 1 n m - JJWy te y)dxdy (2) The gradient allows determining the predominant direction of the defect propagation and the nonunifor-mity degree of the surface investigated.7 A low mean value of the gradient indicates an insignificant variation of intensities along the given axis of an image. In practice this shows a more uniform picture of the deformation relief in a certain direction8 and indicates the coordinate axis that corresponds to the predominant direction of the defect propagation. In order to enhance the informative features corresponding to the elements of the damaged surface, binary transformation was applied to the obtained grayscale images.9 This resulted in black-and-white images of the damaged surface with the intensity function IB, in which white pixels correspond to the background and black ones to the objects of the deformation relief (Figures 1e to g). The most general parameter that allows evaluating the degree of specimen damage using the obtained images, is the relative area of defects: = 5 m-n -•100% (3) where S is the number of deformation relief pixels in an IB image. The distribution of deformation relief elements along the image axes is described with horizontal Hx and vertical Hy histograms:8 Hx (y) = 11(x, y), Hy (x) = 1 I(x, y) x =1 y =1 (4) Each element of a histogram contains a number of pixels that correspond to the objects of the deformation relief, in columns and lines of the image analysed, respectively. Histogram functions (4) contain the basic information about the distribution of the deformation relief along the coordinate axes of an image. For a generalised evaluation of the surface damage based on histograms it is proposed in8 to use the mean values of histograms fi x = S / n and fi y = S / m (where S is the general number of black pixels). However, it is reasonable to use these parameters during multiple measurements under similar conditions with a permanent rectangular watch window. During laboratory testing of different specimens, especially under different conditions of the surface defect nucleation, the mean values of histograms contain little information. In addition, in the case of a rectangular watch window the values of ^ and ^y are scaled differently (relative to the image dimensions) and are inconvenient for a comparison, while in the case of a square watch window they become similar. For a quantitative evaluation of the histogram view (4) a spectral analysis of the functions was performed. Using the fast Fourier transformation the histogram functions were presented in the form of a row: ^^ k H. (y) «i Axk cos(2tt- y -px) k=0 i (5) Hy (x)« i Ayk cos(2n - x-