Simulation of Microstructure and Mechanical Properties in Ductile Iron

Development of Casting Process Simulation for Iron Castings:

Fig. 1: Timeline of simulation tool development

Initially the term "Solidification Simulation" meant exactly that. Tools used a homogeneous temperature distribution (one temperature) throughout the entire casting as starting condition. Very often just single values for thermophysical properties, i.e. density, conductivity, specific heat capacity were considered by the codes, not temperature dependent values. Some tools didn't even consider the mold material surrounding the casting. Those tools were used to predict hotspots in castings. These tools considered neither the influence of the temperature loss of the melt during the filling process, nor material transport phenomena. The use of these tools lead to many over-risered castings especially in iron foundries. The solidification of gray and ductile iron is characterized by the interaction of multiple components and their volume changes. One of the most important factors is the graphite expansion versus the shrinkage of the metal matrix during the iron solidification. Both are influenced tremendously by the metallurgy, i.e. the composition, inoculation, graphite precipitation, and the melt treatment. Not to be forgotten should be the influence of the mold material with regard to mold stability (mold wall movement) and moisture content. 

The need to consider all of these factors leads to the necessity to micro-model the creation of the microstructure during the solidification (Figure 1). Actually, it is beneficial to consider certain effects, like fading of inoculants and pre-solidifying sections of the casting, during the filling process. In many cases sound iron castings can be produced without risers. But only highly sophisticated casting process simulation tools can be used to simulate these kind of casting successfully. 

The modeling of the microstructure during the solidification process allows the tool to continue simulating the cooling process of the casting all the way down to room temperature. Hence, a prediction of the microstructure at room temperature can be made in conjunction with a prediction of mechanical properties. This functionality becomes more and more important for the cooperation between the iron foundries with casting designers, especially in combination with the prediction of residual stresses and distortion in castings.

Example 1: Ductile Iron Ring Casting

The foundry producing the 5600-lbs. ductile iron (Grade 80-60-03) ring casting had problems with under riser shrink. No matter how many risers they used, always shrinkage porosity appeared below the risers (Figures 2 through 4).

Fig. 2: Ring casting with removed risers

Fig. 3: Detail view of broken off riser connection

Fig. 4: Picture of shrinkage under riser

The casting is poured into a very rigid chemically bonded mold, which would allow the foundry to consider a riserless gating design. However, the use of a simple "Solidification Simulation" tool predicted a ring shaped shrink inside the casting (Figure 5).

Fig. 5: Ring-shaped shrinkage predicted by "Solidification Simulation"

At that time the yield of the casting was 77% and the scrape rate was 50%. It was decided to use a casting process simulation tool (MAGMASOFT) to reproduce the present under riser shrink and verify the appropriate process setup in the casting process simulation. The initial casting process simulation considering the filling process, the metallurgy and melt treatment, as well as, the appropriate mold stability and properties reproduced the present under-riser shrink (Figure 6)

Fig. 6: Simulated under-riser shrink

Fig. 7: Simulated under-riser shrink

 second run was conducted using a riserless design. The results show a casting with only minor defects on the surface, but none inside the casting (Figures 8 and 9).

Fig. 8: Minor surface shrink on riserless design

Fig. 9: No shrink inside casting with riserless design

After these simulation runs the foundry started producing defect-free castings without risers. The minor surface shrink, if present, is of no concern because it gets removed by the machining. The financial impact of this change is significant (Figure 10).

Fig. 10: Savings of more than US$ 80,000.00 per year have been realized

Not only are the costs reduced for each new casting due to yield improvement, elimination of exothermic sleeves and riser removal costs, but the overall scrap rate has been reduced to 4%, too. This eliminated the need to produce additional 20 castings per year to deliver 40 sound castings. Using "Casting Process Simulation "instead of" Solidification Simulation saved more than US$ 80,000.00. 

This example proves that it is essential to consider the entire casting process and micro-model the creation of the ductile iron microstructure to get an accurate shrinkage prediction. 

Example 2: Ductile Iron Ring Casting

A ductile iron test casting was poured as part of the Thin Wall Iron Group (TWIG) research program. The part included interconnected plates (stair step casting) and separate plates with different wall thickness ranging from 2 to 6 mm in thickness (Figure 11).

Fig. 11: Test Casting

Image analysis was used to evaluate the microstructure in a center plane of the stepped area and the separate plates. A casting process simulation was conducted considering the entire casting process, the metallurgy and the cooling process including phase changes to predict the as-cast microstructure at room temperature. The comparison of the measured and the simulated values for nodule-count, ferrite and carbide distribution show very close matches (Figures 12 through 16). Besides the confirmation of the wall-thickness dependency of the microstructure it was also confirmed how important it is to consider the temperature loss of the melt during the filling process and the resulting preconditioning of the sand mold due to the filling. Differences in wall-thickness and the resulting differences in local cooling rates, alone cannot explain the measured distributions in the stepped area of the casting. 

Fig. 12: Comparison of nodule-count distribution in step plate

Fig. 13: Comparison of ferrite distribution in step plate

Fig. 14: Comparison of carbide distribution in step plate

Fig. 15: Comparison of nodule-count distribution in separate plate

Fig. 16: Comparison of ferrite distribution in separate plate

Example 3: Mechanical Property Prediction in Ductile Iron Crankshaft

In the frame of a casting engineering project the casting process of a ductile iron crankshaft was evaluated with regard to casting defects, microstructure and as-cast mechanical properties. After implementing the process conditions present in this particular foundry the simulation showed a close match to the microstructure and mechanical properties found in the castings (Figure 17). The final simulation lead to an optimized gating system, which reduced the filling time by 45% and eliminated inclusion problems, found previously in the castings. The riser size was reduced, improving the yield of the casting. A significant cost reduction was achieved by the elimination of chills, after the simulation showed that a defect-free casting with sufficient mechanical properties could be produced without them.

Fig. 17: Mechanical property distribution in ductile iron crankshaft