3D scanning applied in the evaluation of large plastic deformation

Crash test are experimental studies demanded by specialized agencies in order to evaluate the performance of a component (or entire vehicle) when subjected to an impact. The results, often highly destructive, produce large deformations in the product. The use of numerical simulation in initial stages of a project is essential to reduce costs. One difficulty in validating numerical results involves the correlation between the level and the deformation mode of the component, since it is a highly nonlinear simulation in which various parameters can affect the results. The main objective of this study was to propose a methodology to correlate the result of crash tests of a fuel tank with the numerical simulations, using an optical 3D scanner. The results are promising, and the methodology implemented would be used for any products that involve large deformations.


Introduction
Crash tests are fundamental to evaluate the safety and functionality of automotive components. The aim is to simulate car performance in case of accidents, evaluating the results and determining if certain safety standards are met.
In order to regulate car safety standards, regulatory agencies were created to qualify and certify have to meet crash test standards, such as ABNT NBR 15300 (2005), for frontal impact, and ABNT NBR 15240 (2005), for rear impact. • Recyclability -steel tanks are very recyclable.
• Durability -steel tanks produced by Sasft meet stringent requirements for corrosion and structural durability.
• Safety -in the event of frontal, side or rear impact, it is necessary to maintain the integrity of the fuel system without leakage or explosion. Steel, due to its high strength and great capability to absorb impacts, has shown satisfactory performance in all safety tests.
In regard to safety, fuel tank location in the vehicle is a decisive factor. Fuel tanks located on the rear, depending on the impact level, may cause fire in case of rear collision. In pickup trucks and sport utility vehicles, the fuel tank is usually located on the side of the vehicle and oriented lengthwise. In this case, the main risk of explosion is in side collisions.
Besides its location, the design and manufacture of a fuel system in a car has a key role in its security. The tank and the filler pipe must meet certain security requirements in case of possible collisions of the vehicle. Usually, automakers create their own safety standards for specific components, such as the fuel tank. In early stages of design (and in the validation of a component of a new supplier), it is useful to perform the crash test for the fuel tank and then, at a later stage, for the entire vehicle. For the tank, many manufacturers use a sled test (SILVEIRA et al., 2008). This procedure consists of a sled guided by rails, raised to a certain height and then released to crash the tank (partially filled with water) with a pre-established kinetic energy. In this test, deformation and tightness of the tank are evaluated after the impact.
These impact tests are classified as highly destructive tests and generally need to be carried out more than once to completely validate the prod- The aim of this study was to propose a numerical/experimental evaluation methodology of a metal fuel tank that endures significant deformation in crash tests, using a 3D laser scanner. where m is the mass of the sled and v its speed.

Experimental test
At first, ignoring the losses by friction, it can be said that the potential energy (E p ) of the sled at the maximum height is equal to the kinetic energy (E c ) of the sled at the time of impact. Thus, one can determine the height at which the sled should be released to reach the speed equivalent to the specified energy as follows: ( 2) where g is the acceleration of gravity and h is the height.
However, in a real test, there will be considerable losses by friction between moving and fixed parts of the equipment -e.g., between pulleys and rails -which need to be taken into account to calculate the real height h. A device to measure the velocity just before the impact was used to accurately calculate the total loss by friction. After five measurements, an average loss of 10% was found.
The experimental test data are given in Table   1. All the open parts of the tank were sealed to prevent leakage of air or water, creating considerable internal pressure during the test.

3D Laser scanning
In addition to being extremely slow, conventional measuring machines, tactile digitizers and CMMs involve a contact force between touchprobe and measured object. The advantages of optical measurement techniques are evident.
A large number of different techniques for optical measurements are currently available such as: shape from shading, shape from texture, time of flight and light-in-flight , laser scanning, laser tracking, Moiré interferometry, photogrammetry, structured light, etc. (LEMES, 2010).
In this study, a 3D laser scanner was used, that is, a device used to capture in a short time physical objects in a digital format without contact with the object. It works by projecting a la-   In this study, the Updated Lagrangian method was used and the tank modeled with plate elements, bi-linear interpolation and six degrees of freedom per node. In this case, the Mindlin model was adopted, where the crosssection remains plane after deformation but not necessarily orthogonal. This element has only one point of integration in the plane, which is convenient to avoid locking problems, but can generate questionable results due to the hourglass phenomenon (spurious zero energy modes).
To minimize this problem, it was necessary to add "anti-hourglass" forces and moments, using a plate formulation with physical stability.
Along the thickness, five points of integration were used. The material was modeled as elasticplastic with isotropic hardening and the von Mises yield criterion. The update of the stresses is done through an incremental iterative process proposed by Mendelson (1968) that is more accurate (but with greater computational cost) than the traditional radial return.
To model the effects of the fluid inside the tank, a Lagrangian formulation was used based on hydrodynamics particles (IDELSOHN et al., 2004).
And to simulate pressure effects due to volume reduction of the tank during the impact, volume con-  Table 2 (THOMPSON, 2006): The data related to temperature were ignored, since the temperature variation during the impact is insignificant. In previous studies, it was observed that the effect of strain rate is negligible in this sled crash test. In simulations considering the strain rate of the material, the difference in deformation modes was very small. Figure 5 show that the fuel tank was largely deformed after the crash, despite the relatively low impact energy (3200 J). Figure 6 show the graphics of the energy variation (internal, kinetic and hourglass) over time, showing that the hourglass energy was very low, which proves that the "anti-hourglass" formulation inhibited the spurious modes.   It can be seen that the internal pressure increases by approximately 50% due to the reduction of the volume in the tank. The ratio between water (incompressible) and air (compressible) inside the tank is a key factor in the amount of deformation. With the geometries superimposed, it was possible to make cross-sections on the tank and

Conclusion
In this study using optical 3D laser scanning, a methodology was implemented to evaluate numerical/experimental results of components subjected to large deformations. The digitalization of the experimental results allowed evaluating the modes and levels of deformation along the component.
From the results, it will be possible to adjust the parameters of numerical simulation, changing the mesh refinement, types of elements, constitutive models, hardening laws, yield criteria, etc., in a way that will make the crash simulation strongly reliable. This methodology can be applied to other automotive components such as bumpers, doors and safety side bars.
To improve the accuracy of numerical simulation results, data of the stress/strain curves for different strain rates are being studied. The data obtained from the simulation result of the stamping process, such as residual plastic strain and final thickness, will be imported into the numerical model to make the input parameters closer to reality.
For a future studies, using appropriate CAD software, a more accurate correlation will be carried out through a qualitative and quantitative analysis of the measurements of the differences between the numerical and experimental results.