Predicting Spall Strength of Metals and Alloys Using Data Analytics and Machine Learning Techniques

The spall strength of a material is an important property for understanding its response to high-strain-rate deformation and shock-wave compression. A high spall strength is associated with resistance to spall failure, which is critical for applications like light-weight, impact-resistant armors and protective shells to contain explosive devices. Typically, the response of materials to these extreme conditions is studied through expensive and time-consuming destructive experimental setups, for example, plate-on-plate impact experiments performed using gas guns. Searching the full materials space for those with desirable spall strength is a cumbersome task. A data driven, machine learning-based method capable of predicting the spall strength could overcome these disadvantages and be used to quickly and inexpensively screen metals and alloys. This work utilizes the measured spall strengths for metals and alloys, which have been obtained from literature and tabulated for examination through data analytic techniques. The dataset resulting from literature has been curated to only contain spall strength values experimentally obtained through optical interferometry, specifically Photon Doppler Velocimetry (PDV). These spall strength values have then been correlated to other, more easily accessible properties that are expected to control spall behavior, such as yield and tensile strengths, elastic moduli, and other physical material properties – referred to in this work as “proxy properties.” These proxy properties have been used to establish design guidelines, resulting in a predictive model that can classify the spall strength of a variety of metals and alloys.