(Georgia Institute of Technology, 1999-10)
Bandte, Oliver; Mavris, Dimitri N.; DeLaurentis, Daniel A.
A key issue in complex systems design is measuring the 'goodness' of a design, i.e. finding a criterion through which a particular design is determined to be the 'best.' Traditional choices in aerospace systems design, such as performance, cost, revenue, reliability, and safety, individually fail to fully capture the life cycle characteristics of the system. Furthermore, current multi-criteria optimization approaches, addressing this problem, rely on deterministic, thus, complete and known information about the system and the environment it is exposed to. In many cases, this information is not be available at the conceptual or preliminary design phases. Hence, critical decisions made in these phases have to draw from only incomplete or uncertain knowledge. One modeling option is to treat this incomplete information probabilistically, accounting for the fact that certain values may be prominent, while the actual value during operation is unknown. Hence, to account for a multi-criteria as well as a probabilistic approach to systems design, a joint-probabilistic formulation is needed to accurately estimate the probability of satisfying the criteria concurrently. When criteria represent objective/ aspiration functions with corresponding goals, this ?int probability?can also be called viability. The proposed approach to probabilistic, multi-criteria aircraft design, called the Joint Probabilistic Decision Making (JPDM) technique, will facilitate precisely this estimate.
(Georgia Institute of Technology, 1999-01)
Mavris, Dimitri N.; Bandte, Oliver; DeLaurentis, Daniel A.
The present paper outlines a method for probabilistic multi-criteria decision making. Recognizing the limitations of traditional probabilistic methods in accounting for multiple decision criteria in conceptual or preliminary design, this new method combines probabilistic treatment of uncertain information with a multi-criteria decision making technique. The paper describes how the method addresses a need in Multi-Disciplinary Optimization and Analysis as well as the advanced technology selection process in conceptual and preliminary design. The mathematical foundations of a general joint probabilistic formulation are outlined. Two specific functions are introduced that compute the joint probability: the joint empirical distribution function and the joint probability model. The utility of both functions is demonstrated in a proof of concept study for two criteria, applying both functions to a challenging aircraft design problem, the High Speed Civil Transport. This example application addresses two pressing issues: the identification of a feasible design space for a given design concept and the evaluation of viability of a given aircraft design. Finally, the advantages and limitations of the empirical distribution function method as well as the joint probability model are summarized.
(Georgia Institute of Technology, 1998-01)
Mavris, Dimitri N.; DeLaurentis, Daniel A.; Bandte, Oliver; Hale, Mark A.
Within the context of multi-disciplinary aircraft analysis and design, a new approach has been formulated and described which allows for the rapid technical feasibility and economic viability assessment of multi- attribute, multi-constrained designs. The approach, referred to here as Virtual Stochastic Life Cycle Design, facilitates the multi-disciplinary consideration of a system, accounting for life-cycle issues in a stochastic fashion. The life-cycle consideration is deemed essential in order to evaluate the emerging, all encompassing system objective of affordability. The stochastic treatment is employed to account for the knowledge variation/uncertainty that occurs in time through the various phases of design. Variability found in the treatment of assumptions, ambiguous requirements, code fidelity (imprecision), economic uncertainty, and technological risk are all examples of categories of uncertainty that the proposed probabilistic approach can assess. For cases where the problem is over-constrained and a feasible solution is not possible, the proposed method facilitates the identification and provides guidance in the determination of potential barriers which will have to be overcome via the infusion of new technologies. The specific task of examining system feasibility and viability is encapsulated and outlined in a series of easy to follow steps. Finally, the method concludes with a brief description and discussion of proposed decision making techniques to achieve optimal designs with reduced variability. This decision making is achieved through a combined utility theory and Robust Design Simulation approach.