Saturday, July 12, 2014

Engineers discover uncertainty

A recent paper in Science describes simulations and systematic analyses of uncertainty as a way to simplify predictions of material properties such as density, hardness and reactivity for so called “designer materials.” Conventional engineering – in education and in practice – focus too much on deterministic methods of evaluation and design, an artifact of the use of traditional materials that are well understood. But such designs are of little use in the modern world, already replete with ugly structures and weak materials. Additionally, most of the traditional engineering designs could be fully handled by computers.

The next frontier for engineering has been designer materials that could substantially change industry, electronics, energy and drug design. The use of deterministic and antiquated tools of yesteryear have been successful in slowing down innovation for the past several decades. Precision has to be abandoned and faster experiments with approximations have to be embraced. Trends need to be evaluated, hypotheses need to be formed, tested and proved and errors are to be expected. Once variations are accepted, then it is easier to predict, test, refine and retest. These techniques have been in use in many areas and it is time that such ideas entered engineering and materials science.

Materials scientists and engineers, laggards of the technology cornucopia, need to come to the party soon.