Image-based Modeling of Naval Steels

A.B. Geltmacher,¹ G. Spanos,¹ and J.F. Bingert^2
1Materials Science and Technology Division
²Detailed to NRL from LANL

Introduction: Alloy steels will continue to be the main structural material in Navy surface ships and submarines for the foreseeable future because of their relatively low cost, their good combination of mechanical properties, and the existing infrastructure for processing and fabrication. The desire to optimize or improve materials has traditionally required lengthy and expensive experimental programs. The development of a different approach, guided by material modeling and simulation¹ and validated by selected experiments, would enable significant improvements in the design of Naval materials. This approach is an important part of the Navy's Material by Design Grand Challenge. The ultimate goal of this challenge is to provide design engineers with enhanced/optimized alloy compositions and processing procedures to meet performance criteria for Naval structures, within reasonable time frames. Currently, alloy development time lines consist of many years using standard material development techniques because of their reliance on trial-and-error approaches. To significantly reduce this development time, state-of-the-art computational techniques over a wide range of length scales (from first principles to macroscopic) will be applied in conjunction with critical validation experiments. The objective of the research presented here is to use image-based finite-element models to investigate the effect of material microstructure on continuum-level response as part of the larger framework in the Navy Grand Challenge. Image-based modeling consists of using realistic, experimentally measured microstructural features as a basis for mesoscale stress/strain simulations.

Material Selection: Performance criteria for new alloy steels for Naval applications include eliminating magnetic signature to produce better stealth capabilities and improving corrosion resistance to reduce total ownership costs of future Naval structures. It is well known that the addition to steels of certain alloying elements (e.g., nickel and chromium) reduces magnetic signature by transforming the steel to the nonmagnetic austenite phase. Similarly, corrosion resistance is normally increased by the addition of chromium to the steel. AL-6XN, a "superaustenitic" stainless steel, was selected as a model material for this research. This nonmagnetic alloy has good corrosion and strength properties due to its high nickel (~24 wt.%), chromium (~20 wt.%), and molybdenum (~6 wt.%) content. AL-6XN is currently being considered for certain Naval applications and represents a baseline composition for future alloy design exercises. The desire to reduce alloy content for cost and environmental concerns means that materials engineers will need to consider the possibility that stress-induced phase transformations can produce magnetic crystal structures in these "leaner" steels, due to loading events. Thus, image-based mesoscale modeling is needed to determine local stress and strain states generated within material microstructures under various loading conditions. Initiation of failure events such as stress-corrosion cracking, fatigue, and fracture, is also critically dependent on these local stress and strain states.

Experimental Characterization: We have characterized AL-6XN to develop statistically meaningful data sets in support of mesoscale finite-element modeling and to examine structure-property relationships in this stainless steel. The principal interrogation method involves automated electron backscatter diffraction (EBSD) analysis, which provides crystal orientation information, in combination with standard imaging of the microstructure. Figure 1 shows a typical crystal direction map for AL-6XN along with a subset of grains identified for further analysis. In this figure, different colors represent different crystallographic orientations of the atomic lattices within each grain. Analysis of the ESBD data reveals grain misorientation relationships and grain boundary distributions for the microstructure. Special grain boundary types can also be identified, including low-energy coincident site lattice boundaries, and these are delineated by color in the grain subset. The EBSD results also provide direct input for realistic polycrystal image-based models. Relative plastic response of AL-6XN to various loading conditions may also be predicted locally through mapping of the Taylor factor for a given applied deformation gradient and set of slip systems. Figure 2 shows the Taylor factor map for the subset of experimentally determined grains highlighted in Fig. 1. This map is useful for validating simulation results.

FIGURE 1
An experimental crystal orientation map measured using electron backscatter diffraction (EBSD) for AL-6XN. A subset of grains from the large map is highlighted on the right-hand side. The grains are color-coded by the crystal direction normal to the measured surface plane. Grain boundary types are differentiated by color.

Finite-Element Modeling: Image-based finite-element models have been used to determine the location and level of stress incompatibilities at the mesoscale. Drs. Papaconstantopoulos and Mehl of the Center for Computational Materials Science at NRL calculate the elastic constants for use in the models from first-principles and atomistic models. The image-based model results show that regions of large stress and strain gradients can be developed due to neighboring grain misorientations. For example, Fig. 3 shows the generation of high local von Mises stress regions under highly constrained loading conditions for the grain subset highlighted in Fig. 1. Comparison of these regions to the experimental EBSD maps shows the local grain configurations where phase transformations, plasticity, and damage may be nucleated.

Conclusions: A goal of the Navy's Materials by Design Grand Challenge is to develop the framework for computation-based design and selection of materials well into the future (~50 years). An important component of the Grand Challenge is the use of multiple computational techniques to design a material to meet specific property requirements. The computational models must cover a wide range of length scales—from atomistic, through microstructural, to macroscopic scales. The research briefly presented here shows the development of an experimental/computational technique that is useful in linking the microstructural scale to the macroscopic response in two dimensions. Full three-dimensional (3-D) models based on 3-D experimental data will be a future focus at NRL.

[Sponsored by ONR]

Reference

¹ G. Olsen, "Computational Design of Hierarchically Structured Materials," Science 277, 1237-1242 (1997).