Multidisciplinary Design Optimization Framework for the Pre Design Stage

Abstract

Presented is a novel framework for performing flexible computational design studies at preliminary design stage. It incorporates a workflow management device (WMD) and a number of advanced numerical treatments, including multi-objective optimization, sensitivity analysis and uncertainty management with emphasis on design robustness. The WMD enables the designer to build, understand, manipulate and share complex processes and studies. Results obtained after applying the WMD on various test cases, showed a significant reduction of the iterations required for the convergence of the computational system. The tests results also demonstrated the capabilities of the advanced treatments as follows:-The novel procedure for global multi-objective optimization has the unique ability to generate well-distributed Pareto points on both local and global Pareto fronts simultaneously.-The global sensitivity analysis procedure is able to identify input variables whose range of variation does not have significant effect on the objectives and constraints. It was demonstrated that fixing such variables can greatly reduce the computational time while retaining a satisfactory quality of the resulting Pareto front.-The novel derivative-free method for uncertainty propagation, which was proposed for enabling multi-objective robust optimization, delivers a higher accuracy compared to the one based on function linearization, without altering significantly the cost of the single optimization step.The work demonstrated for the first time that such capabilities can be used in a coordinated way to enhance the efficiency of the computational process and the effectiveness of the decision making at preliminary design stage.

Enabling exploration in the conceptual design and optimisation of complex systems

“…The proposed dependency analysis technique captures the relationships from variables to models using the incidence matrix method (IMM) developed in Guenov et al (2006Guenov et al ( , 2010a. This method creates an incidence matrix by using a set of defined rules and stores information relating to how variables link to models in a system.…”

“…(1) Create an incidence matrix, I, by using the methodology defined in Guenov et al (2006Guenov et al ( , 2010a and then transpose I to create the VBIM, that is, VBIM = I T .…”

“…(2) Using the VBIM, create a VBDSM by adapting the method used in Guenov et al (2006Guenov et al ( , 2010a.…”

“…In this case, variables R, L/D, cj, V , Wp, g and rho are the design (decision) variables which are specified as inputs, whereas fe, K, Wto, Wtf , We, Wrf and A are computational outputs. It is important to note that even a single run of a simple system like this may involve multiple numerical iterations because of computational feedbacks, that is, outputs that are fed upstream in the workflow to provide the required input(s) to nonhierarchically decomposable systems of models (also known as strongly connected components (SCC) (Guenov et al 2010a)). Composition and management of computational workflows are outside the scope of this paper and are explained in more detail in Guenov et al (2010b).…”

“…An illustrative example of the methodology is presented in Section 3. The example utilises a reduced, but still representative aircraft test case supplied by an industrial partner (Guenov et al 2010a). The continuous and dashed lines in Figure 2 identify the design scenarios that provide the context for the example.…”