Modeling and Simulation of Sealing Spray Application Using Smoothed Particle Hydrodynamics

Abstract

Multiphase flow simulation using Smoothed Particle Hydrodynamics (SPH)\nhas gained interest during recent years, mostly due to the inherent\nflexibility of the method and the physically rather intuitive\nformulation of extra constitutive equations needed when dealing with\nfor instance non-Newtonian flows. In the work presented here,\nsimulations based on an SPH model implemented in the flow solver\nIBOFlow has been used for simulation of robotic application of sealing\nmaterial on a car body. Application of sealing materials is done in\norder to prevent water leakage into cavities of the body, and to\nreduce noise. In off-line programming of the robots in the automotive\npaintshop it is of great interest to predict shape and appearance of\nsealing material without having to resort to trial and error\nprocedures.\n\nThe flow of sealing material in the air between applicator and target\n(car body) is relatively uncomplicated, as the material mostly moves\nat constant velocity until impact on target. The flow of the material\non the target is however more complex, applied material flows at the\ntarget surface due to inertia, gravity and pressure and in order to\npredict the appearance of the applied material, flow equations for a\nnon-Newtonian fluid with an open surface needs to be solved. The\nsealing material is both thixotropic and viscoelastic; the material is\nshear thinning but needs to be sheared for some time before the\nstructure of the material is broken down. Conversely, the regain of\nstructure of the material, and thereby also the increase of viscosity\nwhen shearing is stopped or reduced, is also connected to a delay\ntime. In the model used, the local viscosity is considered obeying a\nfirst order differential equation where the stationary limit is\ndetermined by a Bingham relation.\n\nThe simulation model was built by comparing simulations and\nexperiments at three different stages of complexity. In the most\nfundamental stage the material properties were determined. Using a\nrotational rheometer, yield stress, plastic viscosity and thixotropy\ntime constant was determined and implemented in the simulation model.\nTo verify the numerical behaviour of the rheology, simulated rheometer\nexperiments were carried out and compared with the physical\nexperiments. In the second stage, simulation of application of sealing\nmaterial with a stationary hollowcone nozzle was carried out. To\nverify the simulations, the resulting thickness, width and shape of\napplied material as a function of time were compared to experiments.\nIn the third stage a moving applicator of the same type was\nconsidered, here thickness width and shape of applied material as a\nfunction of applicator to target distances were compared between\nexperiments and simulation. At all three stages the number of SPH\nparticles, /i.e./ grid points, in the simulations was varied in order\nto verify that the simulations were resolution independent.\n\nResults of the simulations show good agreement between experiments and\nsimulations in all stages using no artificial tuning of the models,\nthat is, all parameters used in the models are based on physical\nconsiderations. Furthermore, simulation time on a desktop computer\nindicate that computational power required for industrially relevant\ncases is not prohibitively large, for the most complex cases in this\nwork simulation time did not exceed six hours.