Non-steady-state ensemble arc behavior has been observed during the Vacuum Arc Remelting (VAR) of 508-mm-diameter ingots of INCONEL 718. The liquid metal flow in the melt pool of a 508-mm ingot during VAR has been simulated under two alternative sets of conditions: (1) a steady-state axisymmetrical arc distribution, as has been typically used in modeling work previously; and (b) a transient asymmetrical arc distribution. Due to the computational requirements, neither mass flux nor solidification were modeled; instead, the pool shape was fixed from measurements from a 508-mm-diameter ingot, and a constant pool wall temperature of 1609 K was used. The transient simulation assumed a localized Gaussian arc whose effective center was located at a distance of 0.1 m from the ingot centerline; this simulation rotated clockwise around the centerline with a period of 36 seconds. The steady-state model was simulated with axisymmetrical distributions of current and power input to the pool top surface calculated by time averaging the transient current and power inputs. The standard k-e solver of ANSYS CFX 5.6 software was used for both simulations. The transient model results suggest that 5 seconds of asymmetrical arc behavior is enough to change the pool from steady state to transient and that, after 30 seconds, the flow is almost fully developed (at least to the accuracy of the model) and dominated by the Lorentz force. Aspects of the model results agree with key features of the melt pool observed during VAR.

No abstract

No abstract

No abstract

The energy flows during vacuum arc remelting (VAR) of a 20-in.-diameter ingot of INCONEL 718 have been investigated experimentally, numerically, and theoretically, and the results are compared and discussed. The temperatures at a number of points on the outer surface of a VAR crucible were measured during a melt. A forward heat-flow model was constructed and the (initially unknown) interior heat flux distribution refined iteratively until the predicted crucible temperatures matched the measurements. The model included radial and vertical heat flow within the crucible and the development of a heated cooling water layer near the outer surface of the crucible. Significantly, it is shown that the temperature difference between the crucible outer surface and the bulk cooling water was not a linear function of the heat flux at the crucible inner surface. It is shown that results from the literature of plasma physics can be used to place bounds upon the partition of energy during VAR. These bounds are combined with the numericallyinferred power distribution within the crucible to estimate the position of the ingot top during the experiment and, hence, the overall energy partition. Side-arcing from the electrode to the crucible is shown to be predicted to transfer more energy to the crucible than has previously been expected. Time variation in the measured crucible outer surface temperature was also investigated as a means to estimate the ingot top position, and the results are compared with those from numerical modeling and plasma physics arguments. It is shown that the two methods are in fairly good agreement, but that they are in contrast with some aspects of results reported previously.