Use Simulation to Analyze Macrosegregation, Hot Tears, Heat Treatment in Steel Castings.
In the past decade, the use of mold filling and solidification simulation to investigate the filling patterns and velocities, temperature distributions and mold layout for steel castings has become common practice for many foundries. The goal has been to prevent or minimize problems such as air and slag entrapment, mold erosion, shrinkage and riser size optimization. As the acceptance of simulation as a foundry tool has increased, however, there has been a corresponding increase in the sophistication of the models used, the amount of information that can be obtained from simulation and the types of processes that can be modeled.
This article examines the capabilities of simulation software as it relates specifically to steel casting. Through practical examples, the article discusses how casting process modeling can investigate and predict natural convection metal flow that causes macrosegregation, hot tear formation, and the resultant casting mechanical properties after heat treatment. This expansion of simulation software's capabilities further allows foundries to predict the success and quality of their casting designs.
A major cause of fluid flow during the solidification of castings is the action of gravity on liquid metal density gradients (higher density liquid metal will sink while lower density liquid will rise). One cause of these density gradients is temperature gradients that arise during cooling and solidification (thermal convection). As warmer and cooler molten metal moves (both in the bulk melt and the interdendritic spaces in the mushy zone), the flow in turn alters the temperature field in the solidifying casting. These effects are critical for heavy steel castings with thick cross sections.
The density of the molten metal also varies with the composition, so that gradients of concentration lead to density gradients. Similar to temperature gradients driving thermal convection, these concentration gradients also can drive fluid flow (solutal convection). In steel, carbon (C)-enriched molten metal is lighter than metals containing less C. On the other hand, copper (Cu)-enriched molten metal is heavier than molten metal containing less Cu. Therefore, C-enriched molten metal tends to rise, while Cu-enriched molten metal tends to sink. In a solidifying casting, the molten metal is enriched in or depleted of all of the alloying and trace elements to differing extents (due to differences in partitioning between the liquid and solid), and the total density gradient will be a complicated combination of the thermal and concentration gradients. The flow due to these gradients is termed "thermosolutal convection."
Partitioning of the elements between the solid and liquid phases at the microscopic solid/liquid interface during alloy solidification leads to the enrichment of the interdendritic liquid with the alloying and trace elements. In steel, only 35% of the carbon and less than 10% of the sulfur present in the molten metal is incorporated into the solid at this microscopic interface. This leads to an increase in the concentration of the molten metal between the dendrite arms and the development of microscopic concentration gradients (microsegregation) in the growing dendrites. Although the dendrites provide resistance to the movement of molten metal and reduce the velocities due to thermosolutal convection, there still can be a significant amount of flow through the dendritic network. This flow redistributes the highly concentrated interdendritic molten metal throughout the casting, leading to the large-scale composition variations known as macrosegregation.
The simulation of macrosegregation is of interest due to the detrimental effects it can have on a casting. When significant under-riser macrosegregation occurs, cracking problems often are experienced during riser removal that require extensive and costly repair. When replacing multiple risers with fewer larger ones to reduce costs, the main drawback usually is increased macrosegregation. With simulation, the feasibility of such changes can be thoroughly investigated prior to actual production. Simulation tools for macrosegregation predictions that consider the simultaneous solution of equations describing mass, momentum and solute conservation in the solid, liquid and interdendritic regions have evolved from the modeling of the solidification of binary alloys in two dimensions to multicomponent alloys such as steel in real casting geometries.
Figure 1 shows results along the centerline of a steel casting simulation. In Fig. la, the predicted temperature and velocity fields in the casting at 60% solidified are presented. The color scale shows solidified regions in blue, regions with temperatures above the liquidus temperature in red, and shades of green for temperatures between liquidus and solidus. The velocity vectors are represented by lines where the length indicates the speed. At this point, the largest velocities are in the riser, where no solidification has taken place. However, the enlarged portion of the figure shows that there still is a significant amount of flow through the dendrites during solidification.
Focusing on velocity vectors, it is difficult to visualize the three-dimensionality of the flow. In Fig. 1b, the paths of massless particles released at different times during simulation and carried with the flow provide another perspective of the flow field. For example, particles released in the "legs" of the casting are transported throughout, while a particle released in the center of the riser remains in the riser. This indicates that the interaction between the flow in the riser and the casting is not strong. The particles released in the riser also show the rising of the warmer molten metal in the riser center and the sinking of molten metal along the riser walls as it cools.
The macrosegregation of carbon in the casting after solidification is illustrated in Fig. 1c. The nominal carbon concentration was 0.23 wt %. The most noticeable feature of the macrosegregation patterns is the band of high-concentration solid (the white and yellow colored areas) stretching from the riser down around the bottom of the hole. Carbon has been carried away from the areas above and to the right of the hole, as well as from between the casting legs, leaving lower concentration solid behind (indicated by the blue colors). It is important to note that the simulation results show the highly concentrated region in the riser also reaches into the top of the casting, indicating the formation of so called under-riser segregation.
Figure 2 shows a comparison between the measured (left) and predicted carbon macrosegregation along the centerline of a much simpler casting. The macrosegregation patterns show good agreement with lower composition solid along the side walls and higher composition solid in the top of the casting and into the riser. The predicted magnitude of segregation is slightly lower than the measured magnitude. Both the measured and predicted macrosegregation patterns for the other elements in the alloy are similar to those for carbon. The magnitude of the macro-segregation for each of the elements is dependent on how strongly they are partitioned at the solid/liquid interface during solidification.
Hot Tear Formation
The majority of scrap in steel foundries results from shrinkage defects, oxide macroinclusions or hot tears. Foundry process simulation of the mold filling and solidification processes has proven a successful method for the optimization of heat transfer and fluid flow to eliminate or reduce shrinkage-related problems in steel castings. Although the problem of hot tearing has been known for many years, and the relationship between melt chemistry, temperature and stress has been investigated, only recently have unified solidification and stress simulations been used to predict hot tearing problems in steel castings.
The fundamental idea behind the analysis of hot tear formation using simulation is to couple stress/strain and heat transfer analyses of the solidifying casting. The temperature fields from the solidification analysis are required as an input to the stress/strain simulation to determine the amount of thermal expansion/contraction in the casting. The stress/strain analysis alone can provide valuable information about the residual stresses and distortions that build up during solidification and the subsequent cooling of the castings. A comparison of strain rate results from the stress/strain analysis with the temperature fields from the solidification analysis can be used to find regions that may be susceptible to hot tearing. When a part of the casting is simultaneously being rapidly stretched (undergoing high rates of strain) and is at or near the solidus temperature, this is a high-risk area for hot tearing.
It is important when performing a hot tear analysis (as well as for a stress/strain analysis in general) to consider the effects of cores and the mold in restricting the shrinkage of the casting during solidification and cooling. In stress/strain calculations, there also are several numerical techniques available to handle the inelastic portion of the total strain. For the analysis of hot tear formation, the choice of the model used to account for plastic strain becomes important. At present, the lack of data on the mechanical behavior and properties of common steel alloys at temperatures near the solidus temperature makes the selection of a plastic strain model difficult.
Figure 3 illustrates a hot tear analysis for a steel valve body. The location where hot tear problems were experienced in the foundry is indicated on the photograph in Fig. 3a. The temperature field in the casting at 70% solidified is shown in Fig. 3b. The color scale is set so that areas above the liquidus are red, areas below the solidus are blue, and shades of green are used for temperatures between liquidus and solidus. As shown, the area where the hot tear develops is at this time either just solidified or slightly above the solidus temperature. The simulated normal strain rate along the axis of the casting between 65-70% solidified is shown in Fig. 3c. The strain rate in the region where the hot tear develops is much higher (indicated by the light colors) than for any other region of the casting.
The high strain rate where the hot tear forms has developed after the portion of the casting to the left of the hot tear has completely solidified. This solidified metal contracts and strengthens as it cools, pulling the two flanges toward the center. The mold sand between the flanges resists this contraction. The result is a high rate of deformation of the casting in the region where the hot tear develops. The combination of this high strain rate with the fact that the material in this region still is hot and weak indicates the appropriate conditions for the development of a hot tear.
The evaluation of the combined results of solidification and stress/ strain simulations of a casting can be time consuming. For this reason, the development of a criteria that combines the results of both analyses to show areas at risk of hot tearing in a single picture has been undertaken. The application of such a "hot tear criteria" to a test casting designed to hot tear is shown in Fig. 4. Observing the geometry in Fig. 4 left, as the cylindrical part of the casting solidifies and contracts, the thicker section on the top of the casting under the riser will remain hot and will be torn apart. A high level of the hot tear criteria in this region, as shown by the light colored areas in Fig. 4 right, indicates the location where hot tears form. The goal in the development of the hot tear criteria is to show which areas are at risk for hot tears, and also to quantify the level of risk. With this quantification, the evaluation of whether layout changes reduce or increase the danger of hot tearing can be invest igated.
To determine microstructure and mechanical property distributions in a heat-treated casting, it is necessary to have knowledge of the local temperature history in the casting during the different stages of heat treating (austenitization, quenching and tempering). In a heat treatment simulation, this is accomplished by using boundary conditions that are dependent on the stage of the heat treatment process. During quenching in water, for example, the water bath temperature, a heat transfer coefficient that takes into account the degree of agitation of the water bath (higher values for more agitation) and the Leidenfrost effect (where at high temperatures a vapor film surrounds the casting due to boiling, reducing heat transfer) are used as boundary conditions. A different heat transfer coefficient must be used to model heating (by both radiation and convection) in a furnace during austenitization or tempering.
With the calculated casting temperature history known, a comparison of simulated cooling rates with critical cooling rates necessary to obtain certain microstructures can be used to determine the microstructure distribution in the casting after quenching. These critical cooling rates are determined by statistical analysis and are dependent on the alloy composition as well as the simulated austenitization times and temperatures. The hardness distribution after quenching also can be related to the simulated austenitization and quenching conditions through regression equations obtained through a statistical analysis of experiments from a large number of industrial steel grades. In a similar manner, the hardness, yield strength and elongation distributions in the casting after tempering can be determined based on the calculated quench microstructure, the tempering temperature histories and the steel composition.
Where large steel castings are concerned, the old rule of thumb of 1 hr/ in. for determining furnace cycle times still applies in many foundries. Using simulation, a more accurate estimation of the cycle time needed to ensure the entire casting has attained the required temperature is possible. The ability to predict hardness values at critical areas of the casting after machining also means that changes in the alloy content and the heat treatment can be investigated to obtain the desired properties at the lowest cost.
Results for the simulation of the heat treatment of a cam gear casting with an outer diameter of 2 m made from an ASTM 4135 steel are shown in Fig. 5. The simulated heat treatment process for this casting was:
* austenitization with the furnace temperature increasing to 1751 F (955C) over 5 hr and held there for an additional 10 hr;
* quenching/cooling in air until the surface temperature was below 598F (315C);
* tempering with the furnace temperature increasing to 1004F (540C) over the first hour and held there for an additional 10 hr.
Simulation was to be used to estimate whether the hardness on machined surfaces would differ significantly from those measured on the casting surface after heat treatment.
The predicted bainite distribution in the casting after quenching/cooling in air is shown at left in Fig. 5. The low cooling rates resulting from air cooling lead to a uniform bainitic structure, except in the center of the thicker sections where a ferritic/pearlitic structure is present. The corresponding hardness distribution after tempering is shown at right in Fig 5. The results show that the target hardness of 240-270 HB should be met. Measurements on the upper surface of the casting were around 250 HB, with the predicted hardness being only slightly higher. The results also indicated only a slight difference in the hardness values on the casting surface and at the machined depth.
A Look to the Future
As seen in the examples, simulation tools have progressed from mold filling and heat conduction solidification analyses to address areas such as macrosegregation, hot tears and heat treatment. With ongoing research and development, it is expected that an even broader range of simulation capabilities will continue to become available. Current areas of interest include combined heat treatment and stress/strain analyses to examine quench cracking problems and stress relief during tempering, lost foam casting of steel and the pouring of thin-walled steel castings.
This article was adapted from a paper (99-90) presented at the 1999 AFS Casting Congress
|Printer friendly Cite/link Email Feedback|
|Date:||May 1, 2000|
|Next Article:||Using Computer Modeling to Optimize casting Processes.|