Multi objective optimization and probabilistic design on
aircraft fuel system

Abstract: The fuel system in an aircraft is the largest and the most important system.
Designing a fuel system puts large demands on the engineers and therefore a
computer software (a quantified morphological matrix) has been developed by
Saab Aerosystems. The software is used as synthesis help during the
conceptual phase and helps the designer to look through different concept
solutions and minimizes the risks that wrong concepts will be chosen.

The main goal for this thesis is to develop and improve the quantified
morphological matrix with a multi objective optimization framework and also
build it out with a probabilistic approach for the different concepts. For
deterministic optimization this was done with the weighted sum method which
makes it possible to investigate different conflicting objectives and also
presents the best possible combinations of the objectives. Using
probabilistic design in multi objective optimization can be a complex
mission and therefore the computer program Crystal Ball is used to be able
to handle this kind of difficulties. The software uses a simulation type
called Monte Carlo to simulate a model and to get useful probabilistic
design analyses.

The results presented show that there exist several different concept
solutions dependent on several factors as altitude and the weighting
factors, which are used for the weighted sum method. With a probabilistic
approach, the conclusion can be drawn that there are no differences of the
concept solutions between the deterministic and the probabilistic
optimizations. The concept solutions will be same regardless of
distributions and the objectives calculated with the Monte Carlo method are
the same as when calculated deterministically as a single objective.