Déformation d'une paroi d'acier sous pression jusqu'à la rupture : essais expérimentaux et simulations numériques

Abstract

The objective of this project is to define a simple finite elements model which evaluates the deformation of a steel wall under pressure up to ductile rupture. The numerical simulation results are required by the quasi-steady state model, which calculates the amount of arc energy that is confined by the tank deformation of the power transformer or shunt reactor. Several experiments tests are carried out using a bench test where a steel wall of 3.5 m2 surface is subjected to a static pressure up to ductile rupture. The pressure, the displacement and the deformation of the steel wall are measured in order to make the comparison with the results of numerical simulation. Thus, these experimental tests are simulated by a detailed finite elements model using the ANSYS ® Mechanical ™ software. First, the large deflection option is activated for this structural analysis by Finite Elements, which means that the rigidity of the structure is adjusted according to the displacement. The mechanical properties of the materials are obtained from uniaxial tensile standard tests and the true stress-strain curve is used for better describing large deflections of the steel wall. The steel wall is Meshed into 8-node hexahedral solid-shell elements with five integrations points in the thickness. The element size must be less than 2 mm for the convergence of the deformation results according to the convergence criteria of Sinclair (2008). This is why the sub-model technique is used at locations where deformation is large, it allows a reasonable calculation time with a small element size. As for the connections of the steel wall, a static friction coefficient of 0.8 is applied to its upper and lower surface. The clamping force applied to the bolts générâtes sufficient frictional force to prevent the sliding of the steel wall. The validation of the FEA is done by comparison with experimental results up to the ductile rupture of the steel wall. The results of the average vertical displacement of the steel wall in function of pressure are underestimated by 6 % by the FEA. Furthermore, the FEA reproduced with good accuracy the membrane effect of the principal strain of the steel wall. It underestimates the principal strain by an average of 1 % with a deviation of plus or minus 14 % compared to the experimental results. However, experimental and numerical results diverge more at rupture point due to high strain gradient across the strain gage in this zone. The good correlation between experimental and numerical results at the rupture point of the steel wall is still confirmed. Indeed, the FEA overestimates the principal strain by an average of 7% with a deviation of plus 23 % and minus 15 % compared with the experimental results. Then, the ductile fracture at ultimate strain criterion proposed is applied to the FEA results to validate the results at the rupture of the steel wall. The numerical results overestimate by an average of 11 % the pressure of the bench test and underestimate by an average of 1 % the vertical displacement of the steel wall at rupture. In conclusion, the structural FEA method and the proposed ductile rupture criterion are good approach to assess with precision the displacement and the strain of the steel wall under pressure up to fracture.

Date	28 Oct 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Van Ngan Lê (Supervisor) & Henri Champliaud (Co-supervisor)