Virtual Casting: Simulation Predicts Copper Grain Structure in Steel Bimetallic Components

Every time an aircraft brake, a heavy-duty bearing, or a high-performance bushing is manufactured, engineers are gambling with the microscopic architecture of the metal inside it. The size, shape, and arrangement of individual crystals—grains—decide whether a component glides smoothly for decades or cracks catastrophically under load. For a class of hybrid materials known as bimetallic castings, in which a tough steel body is fused with a wear-resistant copper alloy layer, that gamble has historically been a matter of trial and error. Now, a team of researchers in China has shown that the dice can be loaded. By coupling two well-established computational techniques into a single predictive framework, they have demonstrated that the grain structure of a copper alloy layer can be forecast with remarkable accuracy before a single drop of molten metal ever touches a mold.

The study, published in the Journal of Materials Science, focuses on Cu-11Sn-2Ni, a tin-nickel bronze bonded to steel through solid-liquid composite casting. In this process, molten copper alloy is poured against or into a solid steel substrate, and the two materials unite as the copper solidifies. The result is a component that combines the strength and toughness of steel with the low friction and excellent wear behavior of bronze—a combination prized in aerospace applications and other demanding fields where performance and lightweight design are equally critical. Yet the process carries a stubborn problem: the relationship between casting parameters, such as pouring temperature and cooling conditions, and the resulting solidification microstructure has remained poorly understood, leaving manufacturers without a reliable way to steer quality from the outset.

To close that gap, researchers led by Guowei Zhang and Yuan Chang of the North University of China in Taiyuan built a coupled cellular automaton-finite element, or CAFE, model and calibrated it against experimental data from real castings. The approach is elegant in its division of labor. The finite element portion of the model solves the physics of heat transfer, tracking how temperature flows through the molten copper and the steel substrate as the assembly cools. The cellular automaton portion then sits on top of that thermal map, simulating what happens at the scale of individual crystals: where new grains nucleate, in which crystallographic directions they grow, and how they compete for space as dendrites—tree-like branching crystals—extend into the surrounding melt. Because dendritic growth kinetics depend on local undercooling, the two modules continuously exchange information, allowing the simulation to capture how thermal history sculpts microstructure.

Before trusting the model, the team verified it. Simulated grain structures were compared directly with micrographs of experimentally produced castings, and the agreement was high enough to validate the model’s effectiveness. That validation matters because the CAFE framework is only as useful as its fidelity to reality; once confirmed, it becomes a virtual laboratory in which dozens of casting scenarios can be tested without the cost, time, and material waste of physical trials. It is precisely this kind of calibrated digital experimentation that is transforming metallurgy from an empirical craft into a quantitative engineering discipline.

With the model in hand, the researchers systematically varied two of the most influential process parameters: the temperature at which the copper alloy is cast and the intensity with which the casting is cooled. The results were striking. When the casting temperature was lowered from 1300 degrees Celsius to 1200 degrees Celsius, the proportion of equiaxed grains—crystals that grow roughly uniformly in all directions, forming a fine, isotropic structure—rose from 81.25 percent to 88.69 percent. At the same time, the average grain area shrank from 0.11 square millimeters to just 0.04 square millimeters. In practical terms, a hundred-degree reduction in pouring temperature produced a markedly finer and more uniform copper layer, the kind of microstructure engineers generally associate with improved mechanical consistency.

Cooling conditions proved even more dramatic. When the team switched from gentle air cooling to more aggressive oil cooling, the proportion of equiaxed grains actually fell, from 96.85 percent to 88.69 percent, but the total number of grains increased significantly and the average grain size plummeted from 0.64 square millimeters to 0.04 square millimeters—a sixteenfold refinement. The physics behind this trade-off is a classic story of competition between nucleation and growth. Faster cooling steepens the temperature gradient and drives the melt below its equilibrium freezing point more quickly, creating a powerful incentive for new crystals to form throughout the liquid rather than allowing a few existing grains to dominate. More nuclei competing for the same volume of metal means each one grows less before impinging on its neighbors, yielding a far finer structure. The slight shift away from fully equiaxed morphologies reflects the columnar growth favored by steep directional gradients, in which elongated crystals race along the heat-flow axis.

Perhaps the most consequential contribution of the study goes beyond cataloging these trends. Through parameter fitting, the researchers distilled their simulation results into a power-law equation linking the average grain size, the maximum surface grain density, and the pouring temperature of the liquid metal. They then constructed a nonlinear response equation connecting structural parameters to process conditions. In effect, they converted a complex, computationally expensive simulation into a compact mathematical tool: a foundry engineer can now plug in a proposed casting temperature and cooling regime and receive a quantitative prediction of the grain structure that will emerge. This transforms microstructure from an outcome to be inspected after the fact into a design variable that can be specified in advance.

The implications ripple outward across the manufacturing landscape. Bimetallic bronze-steel components are workhorses in sliding bearings, bushings, and tribological systems, and prior research has shown that their wear performance and fatigue life are intimately tied to the grain scale of the copper layer. Finer grains typically mean more grain boundaries to impede dislocation motion and crack propagation, translating into better strength and more predictable wear behavior. A validated predictive model therefore does not merely describe microstructure—it provides a lever for tailoring component lifetime. The same CAFE methodology has already proven its worth in predicting grain structures in superalloy castings, aluminum billets, titanium ingots, and steel blooms, and this study extends that track record into the specific and commercially important territory of copper-on-steel composite casting.

There are broader lessons here about the future of materials manufacturing. As industries push for lighter, stronger, and more efficient machines, hybrid materials that combine dissimilar metals will only grow in importance, and the interfaces and microstructures within them will demand the same quantitative rigor that has long been applied to geometry and loads. The work by Zhang, Chang, and their colleagues—supported by the Natural Science Foundation of Shanxi Province and the China Postdoctoral Science Foundation—shows that the barrier between process parameters and microstructural outcomes can be crossed with tools that already exist. The foundry of the near future may look much like today’s, but behind the scenes, every pour will be rehearsed in silicon first, with grains counted and measured before the metal ever melts. In the ancient art of casting, that is nothing short of a revolution.

Subject of Research: Grain structure prediction in Cu-11Sn-2Ni/steel bimetallic castings using coupled cellular automaton-finite element simulation

Article Title: Grain structure prediction of copper layer in Cu-11Sn-2Ni/Steel bimetallic castings using the cellular automaton-finite element method

Article References: Zhang, G., Chang, Y., Zhao, L., Zhang, Z., Shi, Z., Ren, X., Xing, M., Xu, H., & Zhang, D. (2026). Grain structure prediction of copper layer in Cu-11Sn-2Ni/Steel bimetallic castings using the cellular automaton-finite element method. Journal of Materials Science. https://doi.org/10.1007/s10853-026-13882-7

Image Credits: AI Generated

DOI: 10.1007/s10853-026-13882-7

Keywords: bimetallic casting, cellular automaton, finite element method, solidification, grain structure, copper alloy, Cu-11Sn-2Ni, steel, equiaxed grains, cooling rate, casting temperature, computational materials science