The research objective of this EAGER award is to establish the foundations of a method for managing uncertainty within multilevel design processes. Mitigation of uncertainty in designing complex systems typically results in a significant increase in computational cost; efforts to reduce computational effort necessitate the use of simpler models resulting in increased uncertainty. Hence, there is a need to research and develop methods to find the right balance between uncertainty and computational effort while making good design decisions. The central hypothesis is that the effect of uncertainty and its propagation in multilevel design process networks can be managed by balancing the effort of uncertainty mitigation and the benefit gained in terms of system performance. On successful completion this project will provide the foundational knowledge and approaches to managing uncertainty in the multilevel design of complex systems.
If successful, the outcomes of this award will be: a) an approach for quantifying modeling and computational effort; b) an approach for making tradeoff decisions between the effort associated with uncertainty mitigation and design outcomes; c) a framework integrating information economics and robust design; and d) a validated concept that can be used to carry out systematic studies on uncertainty mitigation in simulation-based design of complex engineering systems. The proposed research is transformational because of its focus on uncertainty management through tradeoff between effort and benefit, as opposed to uncertainty mitigation approaches that are characterized solely by the need to increase the accuracy of models in a design process network. Successful demonstration of the proof of concept will result in a foundation upon which more extensive research will be performed and proposed. In addition to publishing in the scholarly literature, the industrially relevant portions of this work will be presented at workshops and trade meetings. The PIs will proactively recruit and mentor underrepresented minorities and women through their activities ASME and SWE student chapters, Pi Tau Sigma, and programs to attract under-represented minorities.
In spite of tremendous strides in computational capabilities, engineering designers continue to demand greater and greater capabilities for the digital simulation and modeling of products. By using simulation innovative ideas can be explored and refined easily and existing products can be modified – all digitally. However, we are still in the early stages of learning how to manage our new digital capabilities effectively. There is the push to model everything precisely and compute with greater and greater accuracy, but this can be time consuming and costly and is not always necessary. In this proposal, in collaboration with our partner, Dr. Jitesh Panchal of Washington State University, we develop a proof of concept for a method of managing the process of engineering design while spending resources most effectively. We combine methods for robust design and information economics to offer designers the option of managing a design process and incorporating refinements on an as-needed basis. In robust design, the object is to create products that are insensitive to variations – variations in input, in environment, and in changes over the product’s life cycle. We extend the paradigm of robust design to propose a range or set of possible products which will be successful regardless of uncertainty embodied in the models used to design them. Of course, there may be a penalty in product performance or cost for this flexibility. The extent of this penalty may be balanced against the anticipated cost of obtaining more accurate information by using information economics. The robust design procedure, the Inductive Design Exploration Method (IDEM), has been combined with the Improvement Potential method which incorporates information economics. This makes it possible to carry out modular computations, rather than the more common all-in-one procedures which is often used. This makes it possible to enhance or refine modules without having to recompile or re-run the complete problem. This method is useful for problem of multiscale design in which information from a smaller length or time scale is passed on to successively larger scales. Problems such as these are especially difficult because typically different communities of researchers simulate the problem on different scales and interactions between these groups is complicated because their expertise may not overlap. Materials design is an important focus for multiscale design. Much of the research in the development of advanced materials has been accomplished experimentally and this is both costly and time consuming. By designing materials and products simultaneously, it is anticipated that both the cost and development time for new materials will be reduced and that materials may be tailored to meet the needs of specific products. Proof of concept for the procedure was demonstrated by the multiscale design of the shell of an underwater autonomous vehicle. For the autonomous underwater vehicle, simulations of the manufacturing process for the material from which the submersible is made, its material properties and the shell structure were integrated. Detailed simulations of the material properties at the different scales have been developed by our colleagues at the Indian Institute of Technology-Kharagpur. These simulations are then successfully used to demonstrate proof of concept of the proposed method combining robust design and information economics. A ranged set of designs satisfying the specifications have been proposed based on integrated information about the manufacturing processes which determine the development of material microstructure and how the microstructure influences material properties and how these properties determine possible configurations for the underwater autonomous vehicle shell. The results have been validated using the validation square procedure. The broader impacts of this effort involved incorporating portions of the research in teaching. Also the work was presented at conferences and has been published in a journal. A further proposal on this work has been funded by the National Science Foundation. Dr. Mistree has worked with students applying for various scholarships/fellowships as well as mentoring junior faculty in proposal writing. He has also been invited to participate in the DOD/NSF study on "Integrated Computational Materials Engineering (ICME): Unlocking the Potential and Realizing the Vision."
