Composites made from E-glass/epoxy or aramid/epoxy are frequently used in aircraft and aerospace industries. These materials are prone to suffer from the presence of delamination, which can reduce severely the performance of aircrafts and even threaten their safety. Since electric conductivity of these composites is rather small, they can propagate electromagnetic waves. Detection of delamination damage can then be monitored by using an electromagnetic penetrating radar scanner, which consists of emitting waves having the form of short time pulses that are centered on a given work frequency. While propagating, these waves undergo partial reflection when running into an obstacle or a material discontinuity. Habitually, the radar is moved at constant speed along a straight path and the reflected signal is processed as a radargram that gives the reflected energy as function of the two-way time and the antenna position.In this work, modeling of electromagnetic wave propagation in composites made from E-glass/epoxy was performed analytically. The electromagnetic wave reflection from a delamination defect was analyzed as function of key intervening factors which include the defect extent and depth, as well as the work frequency. Various simulations were performed and the obtained results have enabled to correlate the reflection pattern image features to the actual delamination defect characteristics which can provide quantification of delamination.
Ground Penetrating Radar (GPR) is largely used nowadays in civil engineering applications as an effective technique of structural monitoring. This paper describes a method using radargram acquired by the GPR for estimating the radius of a steel rebar buried in a concrete massif, which is assumed to be homogeneous and isotopic. Considering the forward problem under B-scan procedure, a closed form mathematical formula was derived for the hyperbola trace appearing as the diffraction pattern provided by a circular steel bar. This equation relates the observed hyperbola parameters in the obtained radargram to the characteristics of the object and those of the host medium where the object is buried. The inverse problem was solved in two steps. Using the extracted raw data in terms of the hyperbola characteristics and Hough transform, the coordinates of the peak hyperbola are identified. Then the wave speed, the rebar radius as well as the coordinates of the rebar centre are estimated by a curve fitting procedure which is based on the selection of an arbitrary set of points on the considered hyperbola. The effect of noise resulting from variations affecting the electromagnetic wave speed was assessed. The noise was assumed to be a random variable and to act additively on the actual ordinates of the hyperbola. The obtained results have shown that the estimation of the rebar radius is very sensitive to the considered level of noise. Noise impedes the retrieval of a bar radius if its magnitude exceeds 5%.
Wind velocity profile varies in the boundary layer according to a complex pattern that depends hugely on the surface roughness and local Reynolds number. The presence of a macroscopic obstacle on the ground surface modifies considerably the flow characteristics of wind speed profile. In this study, the effect on wind speed resulting from local circulatory motion induced by the existence of an obstacle is analyzed in stationary conditions under the assumption of two-dimensional approximation of the problem. Computational fluid dynamics is used to solve the turbulent air flow equations that consist of Navier–Stokes equations coupled to a K-ε turbulence model. A bounded domain having a rectangular form was introduced in order to schematize the atmospheric region containing the obstacle and wind turbine. The boundary conditions at ground surface were fixed by applying a modified wall law. The other boundary conditions included a logarithmic velocity profile at the input, a uniform speed applied on the upper edge of the rectangular domain and a uniform pressure in the outlet area. To solve the obtained equations, Comsol Multiphysics software package was used. The obtained results have shown that the presence of an obstacle has a huge effect on the wind profile pattern and affects largely the extractable power from wind by the wind turbine system.
Considering a stiffened panel made from an elastic homogeneous and isotropic material which suffers a single localized initial geometric imperfection, assessment of the buckling limit state under in-plane uniform axial compression in the direction of stiffeners was performed. Giving a topological configuration of the stiffened plate, focus was aimed at the combined effect resulting from geometrical dimensions and localized defect characteristics. The perfect stiffened plate taken as reference and diverse imperfect stiffened plates suffering a single localized initial geometric defect of the form of a square depression were analyzed in this work. Extensive parametric finite element simulations were performed according to full factorial design of experiment tables that were built on key intervening factors. It was found that the main parameters controlling the buckling stress for the perfect plate are the plate width, then the web height and width, then finally the interaction between plate width and web height. In case of imperfect plates, the most adverse situation was obtained with the defect placed on the intermediate segments of the stiffened plate. A reduction of the buckling stress as low as 56% was reached in this situation. The main factors influencing the buckling load for the imperfect plate differ according to the defect configuration.
Authors:El Kabira El Mjabber, A. El Hajjaji, and A. Khamlichi
Considering the below rated power region of a variable speed wind turbine approximated by a single-degree-of-freedom system, a hybrid adaptive controller based on sliding mode control and radial basis functions neural networks was proposed. Stability of this controller was assessed by using Lyapunov approach. The control algorithm was implemented by means of Matlab/Simulink software package. Comparison with other reference controllers has proven that the proposed controller is relevant as it improves performance in terms of produced electrical energy by the wind turbine system while reducing the amplitude and fluctuations of mechanical loads.
Authors:S. Teidj, A. Khamlichi, A. Driouach, and A. Limam
Detection of cracks in mechanical components as early as possible enables monitoring structural health and scheduling efficiently the maintenance tasks such as replacing the critical parts just in time. Vibration analysis based techniques for crack detection have been largely considered in the framework of beam-like structures. This methodology relies essentially on the observed changes of beam frequencies and mode shapes induced by the presence of damage. In the present work, using an explicit analytical model assessing the effect of a crack on beam strain energy, the beam first resonance frequencies as they depend on a single crack defect characteristics were evaluated. The crack equations were obtained by means of fracture mechanics approach. Variations of the first beam frequencies and modes shapes were then related explicitly to the location and depth of the crack. Measuring the beam frequency changes and monitoring their variations can be used to perform identification of the crack defect parameters by solution of an inverse problem.
Authors:S. Samagassi, A. Khamlichi, A. Driouach, and A. Limam
In this work, reconstruction of pressure time signal rising during a non-punctual impact occurring on an elastic structure has been achieved through using direct Bayesian approach. This was performed by means of posterior distribution of probabilities integrating the likelihood and prior random information. In the case of a noisy linear system for which the densities of probabilities associated to the prior information and noise could be assumed to be Gaussians and mutually independent, a new algorithm consisting of two steps was proposed. The first step works like a Wiener filter action and enables to determine the input pressure mean, while the second step yields evaluations of variability of the input pressure signal around that mean. It was found that the proposed method achieved perfect reconstruction of the original pressure taken at the input of the system.
Authors:N. Touil, A. Khamlichi, P. Dubujet, and A. Jabbouri
Liquefaction potential of soils under the risk of seism is usually assessed by using correlation formulas that are based on field tests and historical earthquakes databases. These correlations depend on the site where they were derived. To use them for other sites where seismic history is not available, further investigation is needed. In this work, a one-dimensional modelling of liquefaction phenomenon is performed by using DeepSoil software. The soil data required for simulations were obtained from field tests consisting of core sampling and cone penetration testing. Using reliability analysis, the probability of liquefaction was estimated for sandy soils located in the Moroccan city of Tangier. The obtained results were found to be close to predictions due to Juang semiempirical approach.
Authors:J. El Bahaoui, L. El Bakkali, and A. Khamlichi
Buckling analysis of axially compressed cylindrical shells having one or two localized initial geometric imperfections was performed by using the finite element method. The imperfections of entering triangular form were assumed to be positioned symmetrically at the mid shell length. The buckling load was assessed in terms of shell aspect ratios, imperfection amplitude and wavelength, and the distance separating the imperfections. The obtained results have shown that amplitude and wavelength have major effects, particularly for short and thin shells. Two interacting imperfections were found to be more severe than a single imperfection, but the distance separating them has small influence.
Authors:F. M. Cabrera, A. Khamlichi, I. Losilla, and I. Hanafi
Considering the increasing prospective regarding the various applications of Poly Ether Ether Ketone reinforced with 30% carbon fibres (PEEK CF30), there is a crucial need to investigate its machinability. In this work, focus is centered on non-linear regression based models that can be built to predict roug hness parameters Ra and Rt associated to turning of PEEK CF30 when using TiN coated cutting tool. Attention was paid to one-variant models that can be proposed to represent the effect of the cutting speed on the surface finish parameters. A broad class of non-linear interpolation models was considered. Their aptness to be used in modelling this particular application was assessed. The identification of the various mathematical models was performed by using experimental results that were obtained from CNC turning of PEEK samples. Based on statistical analysis, all the considered non-linear regression models proved to be highly significant and succeeded to fit adequately the experimental results.