Compared to the conventional design of castings, a substantial qualitative and quantitative improvement in assessing the real performance of cast iron materials can be realized.

The casting process has an essential impact on the creation of the local microstructure of a casting. These local variations in microstructure lead to locally varying mechanical properties. The properties of cast iron castings depend on their geometry and are mainly driven by the graphite morphology, microstructure, and discontinuities. Therefore, the chosen metallurgy and process control are essential parameters for the performance.

The conventional casting design process does not consider the impact of microstructure variations on the fatigue/lifetime performance of castings. The designer lays-out castings based on established standards. These standards assume homogenous properties throughout the entire casting. As a result, only one material dataset is considered in fatigue/lifetime prediction analysis tools.

Local residual stresses and microstructure variations are rarely recognized in lifetime prediction simulations. This leads to underestimating failure risks and failing to utilize the full performance potential of the material. Designers are uncertain how close the expected mechanical properties match the ones found in the real castings. A conservative design approach requires them to apply safety factors, which lead to unnecessarily high weight and resulting costs.

Metalcasting engineers suffer from undesired consequences created by this approach as well: safety factors result in thicker walls, which increase solidification times and usually lead to decreased mechanical properties. They are also tougher to feed and cause increasing residual stresses within the casting during its cooling process.

Coupling casting process simulation with fatigue/lifetime prediction analysis is necessary to unlock the full potential of cast materials. Casting process simulation tools need to provide answers to questions engineers and casting designers have. It must qualitatively and quantitatively describe material and mechanical properties.

Integrating local properties allows the designer to customize the casting design for the specific requirements of an application. The transfer of residual stress distributions caused by the casting process or heat treatment and their consideration as additional load in stress analysis runs is easily accomplished and already standard operating procedure during the development of, for example, cylinder heads. Unfortunately, the recognition of local property variations’ impact on the durability of a cast component is rudimentary. The correlation of local properties and fatigue/lifetime prediction has only been done experimentally for specific components, which makes it difficult to transfer the findings to other parts.

Within a German research project, “MABIFF,” the link between casting process simulation and cyclic material properties was established for different cast iron materials for the first time.

The concept of this research project was to couple the varying microstructures of cast irons (GJS-400 and GJV-450), predicted by casting process simulation, with the lifetime prediction for castings. Experiments were used to derive S–N curves (Woehler curves), which resulted in the development of a closed chain between casting process and the prediction of the final lifetime of a cast component. Local properties driven by the production process of a casting now can be transferred into and considered by lifetime prediction tools.

Microstructure Prediction for Cast Iron
Metallurgy and alloying components have an essential influence on the final microstructure and resulting mechanical properties of a casting. The chemical composition and inclusions, the melt treatment (charge materials, melting method, treatment, and inoculation), as well as the local cooling conditions are of utmost importance. Foundry engineers use these process variables to dial in the desired microstructure (graphite form, ferrite/pearlite ratio) and avoid undesired defects (i.e., porosity or dross) and microstructures (i.e., graphite deformations or chill).

Simulation programs need to be able to predict the kinetics of the creation of the different phases locally during the entire solidification and cooling process. This requires, besides the consideration of alloying elements, the consideration of the inoculation and melt treatment process. The impact of these is usually overlapped with the local cooling conditions within the casting. The calculation of the plain macroscopic solidification and cooling behavior cannot consider these parameters. Microstructure simulation is required to calculate at any time in any location inside the casting, the amount and type of phase created based on the parameters.

Besides the gating and riser system and geometry of the casting, casting process simulation considers the chemical composition, melt treatment, and inoculation, as well as other relevant process parameters. The program utilizes these input parameters and local cooling conditions to calculate the locally available inoculation sites, growth of all phases, impact of segregation to calculate the solidification process and resulting local microstructure, and its properties.

The calculation of all phases during the solidification process allows for the prediction of the final microstructure when the casting is completely solidified. During the following cooling process, the diffusion of alloying elements within the austenite is considered to predict the amount of graphite. The additional consideration of segregation effects of alloying elements allows for the accurate prediction of ferrite and pearlite growth during the eutectoid phase transfer. The calculation of cooling conditions below the eutectoid phase transformation leads to the prediction of phase ratios of the matrix (ferrite/pearlite ratio, pearlite packaging). With this, it is possible to predict the static local mechanical properties over the entire casting.

Experimental Research to Locally Couple Microstructure and Expected Lifetime
The framework of this research project included the evaluation of major microstructure characteristics of a ductile iron (GJS-400) and a compacted graphite iron (GJV-450) and their impact on component lifetime. For GJS-400, the impact of its ferrite/pearlite ratio was evaluated. For GJV-450, its nodularity and pearlite content were evaluated. Both characteristics can be predicted locally with casting process simulation. The consideration of inclusions and oxides was excluded on purpose in this research project.

To realize a complete chain of information, both materials underwent comprehensive cyclic load testing using tension–compression and cyclical bending tests for various microstructures.

In order to develop a representative correlation between microstructure and component lifetime, typical microstructures found in the relevant castings were used in the test pieces. Due to the required sizes of the test pieces, only a limited amount of pieces could be derived from the castings (windmill bedplate/frame, bearing support, and crankcase). Therefore, additional test castings were poured.

The different melts were dialed in by the research partners in a manner to achieve typical variations in nodularity for compacted graphite iron (CGI) found in crankcases. The different wall thicknesses in castings and test castings led to sufficient variation in ferrite/pearlite ratios for the gray iron.

The samples were machined and polished to exclusively evaluate the influence of the microstructure on the lifetime performance. The impact of the casting surface was not part of this research.

The fatigue test runs were performed under tension–compression and cyclical bending conditions. Tension and elongation monitored test runs were performed, as the lifetime prediction analysis uses both concepts.

After failure, the fracture surfaces of the samples and the local microstructure were evaluated using automatic image analysis according to DIN EN ISO 945.

Twenty-two characteristics were evaluated, specifically the number, shape, and size of the graphite particles, as well as the ferrite/pearlite ratio. The chemical composition was considered as well.

The microstructure characteristics found also were used to validate the accuracy of prediction of the casting process simulation tool used in this project (Fig. 3).
Comprehensive variance and regression analysis were used by the project partners to derive S–N curves (Woehler curves) from the correlation between the microstructure distributions found in the samples and castings and the durability test runs.

The predicted and measured durability values for samples in tension–compression runs had an 87% correlation. These findings were implemented into the simulation program. Now it is possible to consider process conditions at any location in a casting when predicting durability values.

The durability values distributions developed in this project provide a significant qualitative and quantitative improvement compared to the conventional method.

This integrated virtual process chain was validated on one ductile iron and one CGI alloy.

Local Durability Prediction in a Crankcase
Two major microstructure characteristics drive the strength of CGI: the shape and size of the graphite particles and the ferrite/pearlite ratio (assuming no defects or gray/white microstructure is present). Predicted nodularity and pearlite content of test castings and an Audi 3,0l V6 TDI CGI crankcase were used for the coupling of local microstructures with durability values and validated by measurements.

The simulation tool was used to predict the nodularity and pearlite content in the critical areas. The program predicted a nodularity of about 20% and a pearlite fraction of more than 90% in the critical areas (Fig. 7). Based on these values, the tool calculated local durability values, which were used in the lifetime analysis program.

The consideration of local S–N curves leads to double the number of cycles until failure compared to using the conventional method. This explains why no castings failed on the test stand.

Durability Prediction of Gray Iron Bearing Support
A durability strength value of 190 MPa was measured for a gray iron bearing support. Conventional methods, which are based on established standards, showed only a value of 140 MPa for this material. Results of fatigue tests for the bearing support are shown in Figure 8 for two different melts.

Besides showing these quantitative differences in values, it clearly defines potentially weak areas. Figure 9 compares the conventional (microstructure independent) and improved prediction of crack initiation areas in the bearing support. Only the microstructure distribution-based durability calculation predicts the correct location of weak spots in the casting.

Outlook
The results of this research project are the first steps toward integrating process and application simulation to predict correct and robust fatigue/lifetime values. One essential goal of the work presented here was to realize a new methodology and show its potential. The local material behavior of many cast alloys is not only defined by their microstructure distribution, but also by local defects. This is especially important for ductile iron materials. Their performance can be significantly reduced by inclusions and dross. The calculation of these effects, so they can be considered in predictive tools, is currently being worked on.

The proposed concept of local durability values offers great potential in the design of safety critical components, like wind energy parts. Figure 10 shows the local durability values for a ductile iron (GJS-400) windmill base frame.

The red dots in Figure 11 on the left side show the local fatigue strengths (351–455 MPa) based on conventional standards. That approach leads to a very conservative design compared to when locally predicted durability values would be used. The conventional approach puts ductile iron to a disadvantage.

Casting buyers and designers have learned over the years that early information regarding expected property distributions in castings is contributing to quality assurance and risk minimization. They are also increasingly utilizing this information as a chance to reduce weight and optimize the performance of castings. A casting design where the application load and weight are optimized is only achievable if the designer can fully unlock the potential of the material. The integration of process and application simulation offers the development of more realistic design rules for castings.

This research presented in this article was originally published online in the International Journal of Metalcasting in 2016. It can be found in the Volume 11 April 2017 Issue 2. https://link.springer.com/journal/volumesAndIssues/40962

Click here to see this story as it appears in the March/April 2017 issue of MCDP.