Analysis and Optimization of a Stopper-Rod Pour
In metal casting, controlling pouring systems is important to the quality of the final product. An efficient method for delivery of metal in high production cast lines is to utilize a stopper rod. However, this method does have features that can cause casting defects. For instance, the metallostatic pressure head, the shape of the rod and the speed and height of its removal from the nozzle all control the level of air entrainment and oxide film creation.
David D. Goettsch of General Motors Power Train and Michael Barkhudarov of Flow Science, Inc. have performed experimental and numerical investigations of a production stopper rod in order to determine a method of minimizing metal damage in a throttled condition.
For the experiment, a clay-graphite A50 crucible was modified to accept a graphite nozzle such that a detailed view of the aluminum metal front in the nozzle could be captured using a 450 kv X-ray source, an image intensifier and a 720x480 pixel, 30 frames per second movie capture card. A programmed robot was used to raise the silicon carbide stopper rod at specific speeds and heights. The material poured was A319-aluminum at a temperature of 715°C with an initial head of 0.165 meters.
The simulation configured in FLOW-3D was assumed axis-symmetric with a vertical axis of symmetry. The top boundary condition was fixed pressure with a head 0.145 meters above the upper edge of the domain. This corresponded with the initial conditions of the experiment. The bottom boundary was set to fixed ambient pressure. The computational model included were gravity, laminar viscosity, adiabatic bubbles, surface tension with an assumed non-wetting condition with contact angle of 160 degrees.
There were seven total cases (Case A-G). Four cases consisted of lifting the rod from fully closed to a height of 0.0123m (Case A), 0.015m (Case B), 0.023m (Case C) and fully out of the crucible (Case D). In the fifth case, the motion consisted of a two-stage stroke. The rod was moved up to 0.029m and then down to 0.016m (Case E). Finally, cases 6 and 7 (Case F and G, respectively) were only performed numerically. Case F consisted of a changing the tip shape of the stopper rod with a conical shape matching the draft angle of the nozzle. Case G consisted of changing the wetting properties of the rod and nozzle surfaces. The ease of making changes such as this is what makes computational modeling an ideal tool.
At times close to steady state, X-ray images were overlaid on the computational results showing a comparison of data (Figure 3). With the exception of Case A, the numerical and experimental data correspond well. The discrepancy in Case A is most likely due to the rod shifting to the left during the experiment.
In Case E (Figure 4), in which the rod is raised to purge the air and then lowered slightly, all air is removed by 0.75 seconds which agrees with the observations in Figure 2. For the first five cases the simulations were all run until a steady-state condition was achieved and they show good correlation with the average flow rate observed from the experiments. Case F (Figure 5) shows no flow separation except a small trapped bubble at the nozzle wall. Case G (Figure 6) represents the most significant change to the flow due to the modification of rod and nozzle walls from a non-wetting to a wetting condition. A concave meniscus is formed during the transient stage, thus preventing separation and bubble formation. In conclusion, experimental and numerical studies of the metal flow in a stopper-rod pouring system were performed over a range of rod heights and stages in order to determine their effect on flow separation and flow rate. In all cases where the rod throttled the nozzle, air pockets existed below the rod and along the nozzle walls. The air pocket sizes diminished as the rod height was increased. Also, the two-stage stroke in Case E successfully purged the bubbles and reduced the steady state flow.
The conical design also reduced the quantity of trapped air. And finally, the change from a non-wetting to a wetting condition completely eliminated the air entrapment and produced a flow rate that was higher by 50%. The use of coatings could generate a wetting condition.