Luís Gustavo Cofani dos Santos

An inverse eigenvalue procedure for damage detection in rods

Abstract This work is related to the damage detection problem in a homogeneous rod and consists of finding the location and severity of damage in the rod. It is shown that a finite element model based on the geometric parameters of the rod can be reconstructed from one eigenpair, that is, one eigenvalue and its respective eigenvector. This reconstruction permits development of an inverse eigenvalue procedure for damage detection. The procedure is sensitive to perturbations for a large order model. The use of overdetermined data decreases sensitivity, although this may lead to a more expensive procedure. Therefore, we use a new technique for filtering the information from a single eigenpair to obtain a realizable and less expensive procedure.

Aircraft Air Inlet Design Optimization via Surrogate-Assisted Evolutionary Computation

In aviation, the performance impact of auxiliary air inlets used for system ventilation is significant. The flow phenomena and consequently the numerical model, is highly non-linear, leading to a compromise between pressure recovery and drag for a given mass flow condition. This work follows a step-by-step approach which highlights the important issues related to solving such complex optimization problem, using surrogate methods coupled to evolutionary algorithms. Its conclusions can be used as a guideline to similar industrial applications.

Neutron activation analysis for chemical characterization of Brazilian oxo-biodegradable plastics

The chemical characterization of oxo-biodegradable plastic bags was performed by neutron activation analysis. The presence of several chemical elements (As, Br, Ca, Co, Cr, Fe, Hf, K, La, Na, Sb, Sc, Ta and Zn) with large variability of mass fractions amongst samples indicates that these plastics receive additives and may have been contaminated during manufacturing process thereby becoming potential environmental pollutants.

Quasi-3D Multi-Step Ice Accretion Simulation

The present study assesses a multi-step procedure to perform quasi-3D ice accretion simulations for a 22.5 min failed ice protection system condition. This procedure is based on a 2D ice accretion code (Lewice 2D LEWINT) to predict the ice growth along a two-dimensional section of a three-dimensional body. The solution of gaseous phase of the external flow is obtained through a commercial RANS solver using the SA turbulence model. The water collection efficiency is obtained using an Eulerian scheme and, for each new time increment on the ice accretion simulation process, the computational mesh is re-generated and the external flow updated using patched mesh concept. The results of the simulated ice shapes for a 3D swept wing for an icing exposure time of 22.5 min (divided in 5 time steps of 270 seconds) are compared to the experimental ice shapes. In addition, results obtained using different numerical approaches are presented as well. These approaches are: Lewice 2D stand alone (potential flow solution for the external flow) for 5 times steps, Lewice 2D stand alone using automatic time step and Aeroicing 2D using a single step. The results achieved for the failed ice protection system case showed good overall agreement in terms of horn’s vertical extent, angle and location, however, ice shape horizontal extent related to the wing leading edge showed to be poorly predicted.