Fast Computer Codes for Aircraft Design

In the first interdisciplinary design loop, the w hole aircraft is investigated. To allow the optimiz­er an investigation of the whole flight mission with a sufficient number of configurations, the individual calculations must be very fast and use only few variables. In more detailed investiga­tions. not all disciplines arc involved ai (he same time and perhaps not all mission points. The aerodynamic code can therefore use more time and variables to become more accurate As a re­sult. available turn around time, variable* involved and accuracy achieved rise from step to step until the ultimate step of the flying aircraft.

As long as design modifications by theoretical predictions arc relevant, turn around times arc needed which allow for many repeated design loops. This is, depending on the step: one hour, one night, one weekend Fast codes arc all codes which allow for turn around times of one hour for pure aerodynamic calculations (with many individual code calls) or one night/ weekend for interdisciplinary tasks.

Very fast codes are closed foimulas for the interdisciplinary investigations. They only need some main geometry parameters as input for global estimation of aircraft performance to allow configuration selection.

Fast codes relay on linearised theory with empirical corrections. They need more geometry parameters to allow for a first aerodynamic design optimisation including volume dis­tribution and a first approximation of tw ist and camber: and they check the aerodynamic predic­tions in the interdisciplinary model.

Both codes calculate (at different accuracy levels) the global aerodynamic coefficients for performance calculations and first flight mechanics estimations. They identify the physical drag contributors and provide a load estimation

As any code used for numerical optimisation, the codes must be robust. This means:

The code should be able to calculate all problems which the optimizer may pose If the code breaks down, this must not stop the design process, but the code should deliver an inacccptahly bad result which is the worse the heavier (lie code crash was. For instance, if negative pressures occur, the result can be a bad value proportional to the detected neg­ative pressure value. This leads an optimizer to solutions, where the code docs not crash If the code is reliable, only those arc interesting solutions. Such eases must be controlled by the design engineer!

Today, most research effort is devoted to highly sophisticated CFD-codcs. These codes are needed and must be unproved furthermore, but for a better and practical interdisciplinary aircraft optimisation, quality and applicability of the simple fast codes must be improved Much more research effort is needed in this direction.