Advancements in thermo-material modeling of direct energy deposition processes.

摘要

The objective of this work is to advance the abilities of finite element models of direct energy deposition (DED) processes. DED uses a laser or electron beam to melt either metallic powder or wire, which is controlled numerically to deposit the material on to a substrate or existant component. Computational models of DED processes are used for two primary reasons: process parametrization and prediction of material properties. For the thermo-mechanical models are completed to determine the distortion and residual stresses that develop due to the high thermal gradients inherent during deposition. With accurate models, investigations can be made to mitigate these negative phenomena by alteration of process parameters including laser power, laser path, and environmental conditions. For the 2nd modeling motivation, thermal models are completed from which the thermal gradients, cooling rates, or solidification rates may be computed, which are the phenomenon which drive microstructural development, which in turn determines the final material properties of the deposited material. While DED technology is mature, having long been used to rapidly clad, repair, or build new components, the modeling of these processes is only now able to achieve the accuracy and speed necessary for industrial implementation.;In this work finite element (FE) models are developed and validated for DED processes. In situ measurements of temperature are taken during the single layer laser cladding of InconelRTM 625625. These are used to develop and validate the application of thermal boundary conditions aiming to improve model accuracy. First, the application of convection boundary conditions is explored. During DED deposition, the gases used to propel the powder and prevent material contamination in the melt pool cause a significant convective cooling. Comparisons are made between various methods of applying convection including using natural convection only, forced convection measured from the lumped capacitance method, convection from an impinging jet heat transfer paper, convection measured by hot film anemometry, and ignoring convection. The importance of applying convection to the evolving free surface was also investigated. It Was shown that using the hot film anemometry values applied to an evolving free surface yielded the most accurate model, with 3-13% error. Secondly, conduction losses due to fixturing during laser cladding processes were investigated. Two InconelRTM 625625 laser cladding experiments were completed, each using identical processes parameters, one which was held in a cantilevered fixture, one bolted directly to the work bench. These represent the minimum and maximum contact area during deposition. In situ measurements of temperature were taken, which were used for the calibration and validation of the subsequent thermal simulations. Though two bodies may be in contact, there is a loss of thermal conduction at their junction due to the microscopic irregularities of the surfaces. The drop in thermal heat transfer is called contact resistance and the effective conductivity through the point of contact is called gap conductance. A method for estimating the maximum gap conductance and application to FE models was developed. Calibration of gap conductance was completed for each experimental case. For the cantlivered clad, it was estimated that merely 2% of the total heat loss occurred through the fixture. Application of the gap conductance model improved the accuracy of the model near the point of contact, but did not greatly affect the remainder of the substrate or clad material. For the work bench bolted clad, it was estimate that between 70-85% of the input heat left via conduction. It was shown that the accuracy of the thermal model for this case was vastly improved by using the gap conductance model, but that the effectiveness of the modeling was limited somewhat by the thermo-mechanical interaction.;With the improved accuracy of the thermal models gained by the preceding advancements in boundary condition handling, the focus was shifted towards capturing the material solidification and ensuing microstructure with the FE model. To this end, single track depositions were performed using 4 sets of laser power and scan speed, for both InconelR 625625 and Ti -- 6Al -- 4V. Type K thermocouples were used for far-field validation of the thermal model while high temperature Type C thermocouples were threaded through the underside of the substrate, to lay flush with the surface. This allowed for the in situ measurement of melt pool temperatures during deposition. Post process, the plates were sectioned and the melt depth was measured. Two modeling techniques were used to improve the accuracy for attaining both of these measured phenomenon: altering the melt temperature specific heat to account for changes in liquid density and altering the melt temperature thermal conductivity to approximate convection within the melt pool.