THIS issue’s feature article, Leading a Revolution in Design, focuses on a concerted effort to develop novel design optimization algorithms, methodologies, and software to enable the rapid design of advanced systems, components, and material architectures. With recent breakthroughs in advanced manufacturing—several of which are being enabled by Laboratory researchers—our ability to manufacture complex components and materials has far outpaced our ability to design. Designers, no longer merely inconvenienced by inefficient trial-and-error design, are nearly incapacitated by the vast space of possible designs now afforded by advanced manufacturing technologies. No systematic methods currently exist to explore this vast design space for systems with significant complexity, especially systems of interest to the National Nuclear Security Administration (NNSA), namely, those exhibiting nonlinear, transient, multiscale, and multiphysics phenomena with uncertain behavior.
Although commercial software has been successfully employed to optimize large-scale engineering problems, such software is predominantly limited to simplistic design parameterizations and optimization metrics and assumptions of linear behavior. Academic research solves problems that are more complex but significantly smaller in scale. The engineering community has yet to address the problem of simultaneous optimization over complex design and response spaces in the type of large-scale high-performance computing (HPC) environment that fully leverages the possibilities afforded by today’s advanced manufacturing breakthroughs. The novel research effort described in the feature article aims to close this gap and provide solutions to NNSA’s critical design problems.
To this end, a team of researchers from Livermore and leading universities is creating a comprehensive design optimization software package called Livermore Design Optimization (LiDO). The team’s goal is to fundamentally transform design by using LiDO to suggest radically novel designs that fulfill all the requirements—such as weight, strength, and stiffness—specified by the user. The process of converging to a single design from a set of requirements is often described as inverse design, which, along with optimization, is one of the most compelling frontiers of computational engineering research today.
Complexity in design optimization arises from two sources: design and physics. The complexity of design involves the intricate shapes and material layouts made possible by advanced manufacturing technology, such as structural composites with intricate morphologies and architectures. Design complexity also refers to the strength, stiffness, and other metrics that an engineer seeks to optimize. Physics complexity comes from the mathematical models used to predict design performance. Such models require the solution of partial differential equations involving complex nonlinearities, transients, multiple scales, multiple physics, and uncertainties. Designers usually iterate through the design space, solving the physics equations using numerical methods. With typical degrees of freedom in design and physics exceeding 1 billion, solutions require efficient algorithms running on HPC systems. However, design iterations are usually guided by the engineer’s intuition, with no quantitative way to ensure that the final design will be optimal. LiDO can revolutionize design by allowing engineeers to automatically converge to an optimal design, factoring in all constraints and solving problems of unprecedented complexity and scale. More importantly, this novel capability will make possible highly nonintuitive designs that were previously unobtainable.
The past few years have seen remarkable progress in applying additive manufacturing to the Laboratory’s core mission of maintaining the nation’s nuclear weapons stockpile. For example, as described in the research highlight, "Beaming" Objects with Volumetric Lithography, researchers have invented a method to use light beams to fabricate three-dimensional polymer structures volumetrically and monolithically in a few seconds, rather than building up layer by layer. The Laboratory is also demonstrating additive manufacturing with polymers, metals, ceramics, semiconductors, and novel combinations of materials, creating new opportunities to solve critical national security issues. Additive manufacturing is also enabling advanced batteries; printed biological tissues; and catalytic reactors to convert greenhouse gases into valuable, long-lived products.
Laboratory researchers continue to add powerful new manufacturing tools to their existing toolset to combine novel shapes, internal structures, and physical properties that were impossible to realize in the past but are now limited only by one’s imagination. With LiDO on the horizon, I look forward to continued advances that will surely bolster nearly every area of Laboratory research and U.S. industry.