Click here to see this story as it appears in the September 2017 issue of Modern Casting

Casting processes are often the most cost-effective means to produce geometrically complex components and offer near net-shape capability. The increased use of lightweight aluminum castings in critical automotive structures such as engines, transmissions, and suspension systems, requires improved material properties, with more reliable and quantifiable performance.  The casting quality and final product performance strongly depends upon alloy composition as well as casting and heat treatment processes.  Alloy composition plays an important role in phase transformation and material physical and mechanical properties.  Casting and heat treatment processes determine the size, quantity and distribution of multi-scale microstructures and defects.  With the development of computational methodologies and, in particular, the rapid advance of computers over the past decades, mathematical modeling and numerical simulation of various metallurgical casting processes has become increasingly popular in the metalcasting industry.

Today, modeling with software to predict and visualize heat transfer and fluid flow events is integral to casting process design.  In the past decade, AFS Corporate Member GM launched virtual casting programs resulting in over 20 patents. The vision of the virtual casting program was to integrate manufacturing processes with component design to produce reliable and high-quality cast structural components with minimum lead time and cost.  The virtual casting program achieved its objectives and demonstrated the benefits of its use in both product design and manufacturing process optimization. The application of the virtual casting systems in OEMs and their suppliers has become increasingly widespread, and the methods are being adapted to other alloy systems and processes. 

Virtual casting includes alloy design/selection, process optimization, multi-scale defect and microstructure simulation, and comprehensive analyses and predictions of component mechanical properties and performance.  Virtual casting is realized through the integration of computational tools with advanced materials models linking multiple length scale physics and metallurgical phenomena.  The virtual casting is based on the principle of integrated computational materials engineering (ICME). 

Overview of a Virtual Casting System
The virtual cast component system at GM for aluminum casting consists of four key modules. They are: casting design, process selection and optimization, casting defect and microstructure prediction, and structural performance evaluation. The casting design module creates an optimized geometry of the casting and gating/riser system based on the machined product geometry, structure characteristics, and performance requirements. The process selection and optimization module provides optimal manufacturing procedures for casting, heat treatment, and machining to ensure quality product with minimum casting defects, residual stress and distortion, as well as manufacturing cost.  The casting defect and microstructure prediction module delivers accurate estimates of casting defects and microstructure constituent distributions in the cast components based on the casting design and process inputs.  The structure performance evaluation module conducts a variety of reliability and durability analyses of the cast component based on probabilistic micromechanics models.  The modules can execute individually or collaboratively.  The virtual cast component development modules interface with commercial software.   These modules divide the component domain into small volumes with calculation points referred to as nodes.

In virtual development of a cast component, an initial product geometry is provided to the system for casting design, process optimization, and performance durability evaluation.  Based on the product geometry and property requirements, the casting design module selects the most economic alloy and casting processes for the product and recommends feasible casting and gating/riser system designs.  The casting process modeling tools (such as mold filling and solidification simulations) are used to further optimize the casting and gating/riser system designs and casting process parameters.  The process optimization module also selects and optimizes the heat treatment and machining processes to minimize residual stresses, distortion, and manufacturing cost. 

For a given casting design and manufacturing process, the multi-scale defect and microstructure module simulates and predicts populations of defects and microstructure constituents in every node of the interested product.  The predicted multi-scale defect and microstructure distributions are exported to the structural performance module to predict nodal-based mechanical properties as well as durability of the product.  If the predicted properties and durability meet the requirements, the optimal product casting is designed.  Otherwise, the initial product geometry may need to be modified, and the manufacturing processes such as casting, heat treatment and machining need to be re-optimized.

Optimal Casting and Gating System Design
In virtual casting, the optimal casting and gating system design is performed with a knowledge database, a graphical user interface, a geometry analyzer, an inference engine, a process simulation module, and an optimization module. The knowledge database contains casting design data and rules.  The graphical user interface accepts as input a product design that is to be manufactured by a casting process.  The geometry analyzer evaluates the initial product design and generates the geometric characteristics of the product to be cast.  The inference engine is adapted to generate casting designs by searching the knowledge database, performing pattern-matching operations, and implementing logical processes.  The proposed casting designs from the inference engine are exported to the process simulation and optimization module.

In the optimized gating system design, the thicknesses of the runner and the connectors from the runner to the risers are significantly decreased to reduce liquid metal free surfaces during mold fill.  The riser shape and sizes are also optimized.  Compared with the baseline design, the casting yield of the optimal design is improved by 15% while the predicted casting defects (oxides and porosity) are reduced by more than 25%.  Most importantly, the development cycle (time to get the optimal solution) is significantly reduced from more than two months to about a week by using virtual casting tools.

Process Modeling and Optimization
Based on the input of the casting model and gating/riser designs from the casting design module, the process modeling and optimization module provides recommendations for optimal and robust manufacturing processes to produce high quality casting products with minimum manufacturing cost.  The process modeling and optimization module consists of a casting evaluation tool, a residual stress evaluation tool, and a machining evaluation tool.  The casting evaluation tool assesses a virtual casting defined by the casting design module and cast through a simulated casting process.  The virtual casting is evaluated for formation of casting defects to determine feasibility of the casting design.  The residual stress evaluation tool simulates the heat treatment process recommended for the virtual casting to predict residual stress levels and potential cracks.  The machining evaluation tool models the machining processes applied to the virtual casting to assess dimensional accuracy and potential cracking of the finish machined product.

The predictions of the residual stresses are in very good agreement with the experimental measurements. The module has been used by product design and manufacturing engineers to resolve casting geometry and heat treatment issues.

Modeling of Multi-Scale Casting Defects and Microstructures
The multi-scale casting defect and microstructure module provides detailed predictions of defect populations and microstructure distributions for the given casting and gating design under certain process conditions.  The casting defects that can be predicted in the virtual cast component development system include macro and micro porosity, oxides and inclusions, core gas, cold shuts, entrained air, and hot tearing.  The microstructure constituents that are simulated in the virtual cast component development system include dendritic grains, dendrite cells, and second phase particles in both micro and nano scales.

As an example, the microporosity due to hydrogen, oxides, and shrinkage in aluminum castings, is predicted with the integrated interdendritic flow and pore growth model. The theoretical basis for the pore growth model is that pore growth is governed by the rate at which hydrogen diffuses to the pore/liquid interface. 

The contribution of core gas, created by release of water vapor and CO2 from polymer binder combustion in sand cores and molds, to microporosity can also be modeled as an external hydrogen source.   The concentration can be explicitly stated as a boundary condition, or numerically calculated from formation and transport kinetics for hydrogen and water in the air space in the sand. 

Aluminum oxides, particularly bifilms, provide substrates for microporosity formation. To simulate aluminum oxide distribution in the casting cavity after mold fill, both a scalar variable method and a discrete particle approach were developed. The scalar variable approach is easy to program, but the discrete particle model offers the advantage of directly modeling differential motion between oxides and the liquid metal (including buoyancy). A comparison of the simulation results with X-ray scans of the three castings was interesting. The center casting was of good quality, except in the upper half where macro shrinkage porosity developed. However, both side castings showed that pores also nucleated to the sides just above the ingates. The locations of these pores correlated well with the final locations of the simulated oxide particles. In addition, most of the oxide particles generated during the simulation were found in the two side castings, with a smaller number of particles found in the higher quality center casting.

Experimental validation proved the maximum pore sizes and distribution can be well predicted using the integrated interdendritic flow and pore growth model.  Together with other casting defects and multi-scale microstructure constituents, the predicted maximum pore sizes are used for product performance analysis and product quality improvement.

Prediction of Local Mechanical Properties and Structural Performance
Mechanical properties, in particular fatigue resistance, of aluminum castings strongly depend upon the size and spatial distribution of casting defects and characteristics of microstructural constituents.  The presence of casting defects can significantly reduce fatigue life. However, in the absence of casting defects or when the defect size is smaller than a critical size, crack initiation occurs at the fatigue-sensitive microstructural constituents. Cracking and debonding of large silicon (Si) and Fe-rich intermetallic particles and crystallographic shearing that forms persistent slip bands in the aluminum matrix play an important role in crack initiation. 

With the quantitative prediction of multi-scale defects and microstructures, local mechanical properties of the casting component and its performance under a testing or service load can be predicted using multi-scale fatigue models.

Aluminum shape castings have been increasingly used in structural components because of their high strength to weight ratio, particularly in the heat-treated condition. The mechanical properties and performance of aluminum castings strongly depend upon their microstructure constituents and particularly cast defect population and distribution. To produce quality castings with quantifiable properties, it is essential to control multi-scale defects and microstructure in the casting which are determined by the alloy composition, casting and heat treatment processes.

The ICME tools have been developed and demonstrated successfully at GM for processing aluminum shape castings with desired performance.  The ICME approach and methodology is applicable to other metal and plastic forming processes.  

This article is based on paper 17-017 that was presented at the 2017 Metalcasting Congress.