UC Berkeley

Turbine technology is a critical part of the today's energy and transportation infrastructures, and turbine manufacturers constantly aim to increase the operating efficiencies of their products to yield better technical and economic performance. The thermodynamic efficiency of a turbine system is bounded by its operating temperature, which is often limited by the materials used to construct the turbines blades. The blade materials need to withstand very demanding conditions, including rotational loads from rotations of over 3,000 rpm and thermal loads from temperatures above 1,000 C. Informed materials design of turbine blades is therefore crucial for enhancing the performance of turbine systems.

Ceramic Matrix Composite (CMC) materials are a new series of material systems that have recently received substantial interest from turbine manufacturers due to their exceptional ability to retain their mechanical properties at significantly higher temperatures than commercial superalloy compounds. Given the high cost of experimental CMC research, computation can be a very helpful tool to aid the design of novel CMC compounds by allowing the designer to test the performance of large sets of material systems quickly and effectively through targeted simulations.

This study presents a computational framework that can be applied for exploratory design of CMC compounds. The design framework combines continuum mechanics based simulation tools, such as the finite element method, with evolutionary algorithms to enable computer driven design of CMC compounds. The evolutionary algorithms, driven by numerical and reduced order models, prototype various material and microstructure design choices and evaluate their performance against a set of target properties. As the evolutionary algorithms progress in their parameter choices, they create better and better material design choices until a stopping criterion is reached. Following the down-selection from the evolutionary algorithms, the computer-generated CMC design are then subjected to further virtual tests, including preliminary structural scale finite element simulations of a turbine blade geometry composed of the algorithmically designed CMC compound.