Scholarlink

<p>Realizing composite materials with desired performance requires tailoring their microstructural characteristics to meet ‘many’ (more than three) property goals. Tailoring the microstructural characteristics is challenging owing to the ‘conflicting’ nature of the goals and ‘uncertainties’ associated with the microstructural characteristics. Optimization-based design approaches that assume completeness and accuracy of models and availability of required information are commonly employed to realize composite materials with desired performance by identifying a single-point optimum solution. However, complete information is unavailable during the early stages of design, and uncertainties exist in the models and design variables. Moreover, existing design approaches are often limited to considering a maximum of three goals. Hence, existing approaches are poorly suited for supporting composite material design under uncertainty and highlight the need for robust design exploration to help identify a set of solutions that ensure desired performance under uncertainties. A framework for robust design exploration grounded in the ‘satisficing’ paradigm is presented in this paper to address the limitations of existing design approaches. In the framework, the compromise Decision Support Problem construct is integrated with the robust design metric - Design Capability Index, and a machine learning-based visualization technique – interpretable Self-Organizing Maps. Using the framework, designers are able to (i) generate many solutions that satisfy the many conflicting composite property goals and constraints, (ii) efficiently visualize the associated high-dimensional solution spaces, and (iii) systematically explore the solutions spaces to identify ‘robust satisficing solutions’ that meet the designer’s requirements for the many conflicting goals and are relatively insensitive to the uncertainties in the microstructural characteristics, which was not possible previously. The framework’s utility in supporting designers to tailor the composite microstructural characteristics to realize many conflicting property goals under uncertainty is demonstrated using a sandwich composite material example problem, considering four conflicting performance goals.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Framework for systematic and efficient high-dimensional robust design exploration – a composite materials example

  • 期刊论文

  • 发表日期：

    2025年6月29日

查看全文
  • Niharika Balaji,
  • Mathew Baby,
  • Gehendra Sharma,
  • Palaniappan Ramu,
  • Anand Balu Nellippallil

摘要

Realizing composite materials with desired performance requires tailoring their microstructural characteristics to meet ‘many’ (more than three) property goals. Tailoring the microstructural characteristics is challenging owing to the ‘conflicting’ nature of the goals and ‘uncertainties’ associated with the microstructural characteristics. Optimization-based design approaches that assume completeness and accuracy of models and availability of required information are commonly employed to realize composite materials with desired performance by identifying a single-point optimum solution. However, complete information is unavailable during the early stages of design, and uncertainties exist in the models and design variables. Moreover, existing design approaches are often limited to considering a maximum of three goals. Hence, existing approaches are poorly suited for supporting composite material design under uncertainty and highlight the need for robust design exploration to help identify a set of solutions that ensure desired performance under uncertainties. A framework for robust design exploration grounded in the ‘satisficing’ paradigm is presented in this paper to address the limitations of existing design approaches. In the framework, the compromise Decision Support Problem construct is integrated with the robust design metric - Design Capability Index, and a machine learning-based visualization technique – interpretable Self-Organizing Maps. Using the framework, designers are able to (i) generate many solutions that satisfy the many conflicting composite property goals and constraints, (ii) efficiently visualize the associated high-dimensional solution spaces, and (iii) systematically explore the solutions spaces to identify ‘robust satisficing solutions’ that meet the designer’s requirements for the many conflicting goals and are relatively insensitive to the uncertainties in the microstructural characteristics, which was not possible previously. The framework’s utility in supporting designers to tailor the composite microstructural characteristics to realize many conflicting property goals under uncertainty is demonstrated using a sandwich composite material example problem, considering four conflicting performance goals.