A Look Back at the 20th century...

The 20th century proved to be the most revolutionary in metalcasting's history. With the dawn of the new millennium and as a tribute to the people and technological innovations that have shaped our industry's past, present and future, modern casting is taking a look back at some of the most important contributions of the century.

Casting Process Simulation

While casting process simulation has been available for several decades, only since the 1980s has everyday industrial application of computer-based simulation become common. The ideal combination of software, hardware and personnel has begun to emerge in the foundry industry to allow a huge step in casting production.

Process simulation has been applied to casting solidification, mold filling, residual stresses, microstructure and porosity predictions, and mechanical properties. In addition, simulation techniques, which have advanced to successfully suit the majority of casting processes, have become trusted design tools for foundries looking to:

* improve yield;

* reduce the time from blueprint to production;

* create a slag- and dross-free gating system design;

* shorten lead times and reduce manufacturing costs;

* reduce defects related to mold filling and solidification shrinkage;

* achieve a desired microstructure of the finished product.

Though early simulation applications were restricted to small, simple models, modern high-speed computers have moved process simulation to the forefront of foundry technology. Over the years, the common philosophy in the development of casting process simulation has been making the casting right the first time.

Simulation Roots

Process simulation stems back to the 1940s, when studies on predictions of heat transfer patterns were being conducted. The main application of solidification modeling traditionally was calculation of the path of the isotherms (lines of constant temperature) through shaped castings. In turn, this was used to predict the locations of hot spots in castings and thus to check, using the analog computer, a proposed gating and risering system, rather than following the classical trial-and-error technique.

With the publication of studies using the analog computer, the interest and application of simulation techniques grew. The analog technique was based on the similarity between the differential equations describing the principles of heat and electricity flow.

One of the original analog computers built in the U.S. was the "Heat and Mass Flow Analyzer" at Columbia Univ., New York. Using this unit, Victor Paschkis performed heat transfer analyses on sandcast iron in 1944, several of which were included in AFS Transactions.

The application of this device, according to John Berry, Mississippi State Univ., was confined to a limited number of large metal-casting organizations around the world and was not held in high esteem by foundrymen due to the time required to set up the problem of the analog or to physically determine and connect all the resistors and capacitors required to simulate the cast metal and mold materials.

In 1954, Sarjant and Slack calculated the internal temperature distribution in steel ingots, using numerical techniques to calculate transient two-dimensional heat flow patterns.

Since the main drawback of numerical and even graphical methods was the amount of labor involved in calculating and recording the temperatures at each of the chosen time steps, it appeared natural that the numerical technique could be combined with the abilities of the then modern-day digital computer. This combination of techniques would allow one to solve transient heat transfer problems with unprecedented versatility, speed and accuracy.

Going Digital

According to Berry, the first published reference to the use of the digital computer in a foundry-related application was that of Fursund in Denmark (1962), who studied the diffusion of heat into foundry sands as it affected surface finish problems in steel castings.

Application to the prediction of solidification patterns appeared in 1965, when using the Transient Heat Transfer (THT) digital computer program, developed in 1959 by Campbell and Vollenweider, General Electric Co. (GE), James Henzel and Jack Keverian, GE, discussed the successful application of a finite-difference method (FDM) to heavy steel casting production. FDM refers to the numerical analysis equations that are formed by expanding the partial differential equations using Taylor's Expansion Theorem. FDM's major limitation was the difficulty in tracking the free surfaces of the melt when filling the mold. In the early 1980s, a technique called volume of fluid was introduced by Hirt and Nicholas, in which the fluid fraction function was the main parameter used to track the free surfaces of flow.

The THT computer had high-speed versatility capable of accommodating all thermal and material variables that determined the soundness of a steel casting and yielded the temperature distribution at a maximum of 200 points in one, two or three dimensions for homogenous or composite bodies. The program's reliability was proven on castings weighing as much as 20,000 lb for foundries within GE.

The finite element method (FEM) (Fig. 1) originally was developed for solving complex stress analysis problems for aircraft structures. During the 1960s, the applicability of FEM to both steady state and transient heat transfer problems was reported. According to W.C. Erickson, Los Alamos Scientific Laboratory, Univ. of California, because the basic elements could be defined independent of thermal properties, the problem could be modeled based on information required without being concerned about relative thermal properties.

During the late 1960s, the AFS Heat Transfer Committee sought to encourage further research in the area of computer applications. Sponsorship of research at the Univ. of Michigan resulted in several important publications. Sand castings simulated by Robert Pehlke and James Wilkes were poured in a variety of materials (carbon steel, aluminum and leaded brasses). It was concluded that the predictions of the FDM program were strictly limited by the thermophysical properties of the molding and casting media.

Substantial Gains

With the changing technological atmosphere in the foundry industry and with the groundwork laid prior to 1970, computer casting process simulation had reached a point where its application to industrial problems could be strongly considered.

In his 1971 AFS Charles Edgar Hoyt Memorial Lecture, Ronald Ruddle, Foseco, Inc., made the following prediction: "...we shall within 2 decades see the digital computer take over much of the risering practice." He went on to state, "...it seems likely that in a short while, all one may need to do is feed into the computer a drawing of the casting plus indication of the thermal properties of the mold material and of the alloy being used, its solidification characteristics and the degree of soundness desired in the casting. The computer would then indicate just what would be the most economic risering and gating system to develop the desired degree of soundness and the location of any residual soundness."

In 1973, contemporary with past research on FDM, the Norwegian group of Victor Davies and others extended their FDM work on aluminum sand cast bars to cover permanent mold casting and low-pressure diecasting.

All of the techniques used to simulate the solidification process through the early 1970s lacked the capability to generate the visual representations of the temperature gradients throughout the entire casting-mold cross-section. In 1974, though, the Los Alamos Scientific Laboratory developed techniques whereby color motion pictures could be computer-generated (Fig. 2). This technique was employed in the lab's development of a simulator for the uranium casting procedure. The system used standard microfilm equipment, which was modified by placing a series of color filters between the camera lens and the cathode ray tube and designing control logic for these filters. To describe the various temperature ranges, 11 color combinations were used, which were generated by using the three complementary color filters (magenta, yellow and cyan) and a maximum of 2 strikes/spot.

More Than Solidification

A common failure of many early computer applications to design engineering lay in the inability of the software to summarize the meaning of the many lines of data emerging as output. This led to the appearance of efficient post-processing software capable of displaying, for example, two- or three-dimensional stress plots or isotherms, depending on the nature of the numerical analysis.

According to Nagasaka, Kiguchi and Nachi, Komatsu, Ltd., and the late Keith Brimacombe, Centre for Metallurgical Process Engineering, Univ. of British Columbia, up until the early 1980s, research had been based on the assumption of simulating two-dimensional heat transfer problems.

However, because the shape of most castings is complex, researchers found it important to consider three-dimensional heat flow in the metal and mold in order to simulate solidification accurately. In addition to the prediction of the temperature field and solidification front, they also found it important to develop a method to compute shrinkage cavity in castings that was cost-efficient.

In the 1970s and 80s, as computing power increased and programs were allowed to handle more elements, academic institutions that had spent several years conducting research on simulation began to diffuse into private software houses. These models were able to incorporate fluid flow into the simulation to forecast temperature loss in pouring and predict flow patterns, such as the turbulence in aluminum pours. More importantly, as opposed to systems of the past that incorporated dedicated geometries to a foundry's specific castings, these systems could be applied to any given casting based on enmeshments entered, said Pehlke.

Some of the first foundries to install such simulation packages included Aero Metals, Inc., LaPorte, Indiana; Union Special Corp., Huntley, Illinois; Lesueur, Inc., Lesueur, Minnesota; Haley Industries, Ltd., Haley, Ontario, Canada; and Ford Motor Co., Cleveland.

With software programs capable of visual representations, a second, separate growth period occurred in the 1980s that involved the application of minicomputer-supported, first-generation or subsequent computer-aided design systems. Initially, these systems, which required foundries to undergo serious changes, including the recruitment of a computer-literate/technical expert, were installed for use in connection with rigging design.

According to Erickson, in the early 1980s, the standard assumption of instantaneous mold filling was made. The validity of this assumption depended on the system being simulated. For processes where the filling time was a small fraction of the total solidification time for the casting, this assumption did not introduce significant errors. However, this was not always true in processes where rapid solidification occurred during mold filling. It was the latter areas where fluid flow simulation would play a major role.

Mold filling simulation hit its stride in the late 1980s, allowing foundries to optimize gating systems, eliminate defects caused by filling, determine filling times, create optimal temperature distributions for solidification and feeding, improve yields and effectively use feeding aids and chills.

In the mid-1990s, microstructure simulation was developed and brought with it an understanding of metallurgy effects and the prediction and controlling of mechanical properties. Soon after, convection and segregation simulation allowed users to understand metal flow through growing dendrites. In the late 1990s, stress and distortion simulation introduced the benefits of controlling casting distortion, reducing residual stresses, eliminating hot tears and cracking, minimizing mold distortion and increasing mold life.

Casting Simulation Use

According to the results of a survey conducted by the AFS Process Modeling and Design Committee on simulation use in 100 aluminum foundries, 30% of small shops (less than 200 employees) use it; 80% of mid-size shops (200-500 workers) have installed systems, and all of the large foundries (more than 500 employees) use simulation software.

Simulation software use has been increasing, though, among various metalcasters.

For SivyerSteel Corp., a sand casting foundry in Betterndorf, Iowa, the installation of a process simulation system in 1992 not only advanced its casting knowledge but created new business. For several years, the foundry followed the traditional trial-and-error approach, which would lead to the over-risering of parts and low yield. The system's ability to predict and eliminate inherent sand defects and shrinkage created the opportunity for the foundry to move into new markets.

"It allowed us to produce castings that have a higher selling price," said Jim Hickie, manager of foundry operations. "For us and the entire steel casting industry, a major advantage simulation has provided is the ability to actually see the casting process and be able to gain control of it."

A number of foundries have discovered the benefits that can be realized by installing a simulation program. In the September 1996 modem casting article, "Survey Provides Profile of Casting Design Software Use," a survey conducted by the Univ. of Iowa found a 40-47% improvement (decreases in required time or costs and increases in yield) in part/tooling design and construction lead times among its 154 foundry respondents. Average reductions of 30% were realized in production time, labor costs and rework costs, and an average of 25% improvement was found in the casting yield among the PC users.

The Here and Now

In the early stages of casting process simulation, the foundry industry was hesitant to embrace the technology--only larger, more profitable companies could afford the expense of a system, said Pehlke. Today, simulation software has become a necessity in both small and large foundries' casting design departments, and several commercial packages are available with investment in development of these packages still growing. Further, the fact that non-foundry organizations already use casting simulation in-house means that the level of interaction and understanding between component designers and methods engineers will increase, according to Kevin Selby, The Casting Design Centre.

"The adoption of process simulation has been a question of the vision and resources of the company coupled with a cultural change," said Pehlke.

Today, multiple CPUs can be combined into a single machine and run several times faster than traditional simulations. Where casting designers once waited for simulations to be finished, the systems now wait for them to finish analysis of the result.

Though substantial advances have been made in casting process simulation over the years, several areas of concern still plague researchers. According to Berry, much work still must be done to discover the true nature of the thermal contact between molten metal and the mold as well as a proper description of nucleation within castings themselves.

In regard to future software, Mark Jolly, Univ. of Birmingham, identified that it must be able to predict the occurrence and location of all the problems that occur during the casting process including gas porosity, shrinkage porosity, cold laps and shuts, sand erosion and hot tearing. Second, he said, it should be able to propose solutions to the problems that take into account all of the manufacturing details of each individual foundry.