In general, an engineering system optimization problem is multidimensional, multicriteria, and multidisciplinary. Unfortunately, currently there is no universal methodology for solving such complicated problems. The large number of published papers in the field of multidisciplinary optimization (MDO), and also our considerable practical experience, allowed us to formulate our MDO concept.

Our MDO concept is based on the simultaneous use of multilevel, multicriteria, and parallel optimization technologies that we developed.

A general view of such a concept is presented in Figure below. The attempt to take into account all possible optimization problem statements resulted in the scheme where a system under study is shown as a certain cube with a large number of layers, each corresponding to a separate engineering discipline. Within the frameworks of each discipline, system analysis can be performed using different tools from the simplest mathematical models to possible prototype experimental evaluations.

Our MDO approach is based on the response surface technique and on the original approximation concept. Within the framework of these concept we adaptively use global and middle-range approximations. One of the advantages of the proposed approach is the possibility of ensuring good approximating properties using minimum amount of available information. This is also ensured by application of self-organization and evolutionary modeling concepts. During the optimization process, the approximation function structure is evolutionarily changed, so that we are able to successfully approximate the objective functions and constraints of complicated topology.

The obtained response surface functions can be naturally used by multilevel optimization procedures with the adaptive change of simulation level for both single and multi-disciplinary analysis. These response surface could also be used for interconnection problems solution when solving multidisciplinary optimization problems.

Solution of multicriteria optimization problems is based on the use of response surfaces for individual objectives and constraints. Another part of our strategy is adaptive change of the current search area. This adaptive change makes it possible to search for the Pareto-optimal set without applying convolution approaches. The use of multicriteria optimization algorithms allows us to integrate different disciplines into the united computational process. The given approach is the main one for overcoming the so-called organizing difficulties in MDO traditional concepts.

One of the problem that inevitably arises while optimizing the aircraft compressor, is to ensure high performance not only for one condition (design point), but over the entire operating range. The variables, characterizing a compressor can be divided into two groups. The first group of variables (design variables) contains the characteristics that do not vary when the operating mode is changed. The second group comprises the control variables which can be purposefully changed. The presence of the functional variables that can be purposefully varied suggests the solution of an optimal control problem.

1. Optimum design problem. The goal of this problem is to determine the optimal design variables; the values of these variables are independent from the operating mode.
2. Optimum control problem. The goal of this problem is for the given and immutable variables of the first group find the optimal control law for the variables of the second group for all operating modes.
3. Optimum design-and-control problem. The goal of this problem is to simultaneously determine the optimal values of the design variables and the control laws.