Advancing the materials science of concrete with supercomputers
Jeffrey W. Bullard, Edward J. Garboczi, William L. George, Nicos Martys, Steven G. Satterfield, Judith E. Terrill
Supercomputers are renowned for being used on grand challenge problems like global weather patterns, nuclear device virtual testing, galaxy formation, unraveling molecular structure and now concrete! Why do the mysteries of concrete form this kind of a grand challenge-type problem? Concrete is arguably, from a materials science point of view, the most complex material produced by humankind with distinct and important and random structural features on length scales ranging from nanometers to hundreds of millimeters. It has many mineralogical phases, its behavior is time dependent, it is constituted from natural materials whose properties are quite variable, and it is sensitive to its surrounding environment, which of course drastically varies around the world. Grand challenge! Two of the most important problems for concrete are how its workability or rheology depends on its constituents (placement) and how cement hydration builds up the structure on many length scales (curing and properties). We have made progress on these two problems by using massively parallel computers that drive advanced scientific models. This work has required both sophisticated physical science and sophisticated computer science.