# Present Method Issues

For successful optimization of a parameter space we have to make the objective function topog­raphy suitable for this purpose. This process is described in reference (372j. In this study we used a derivative version of the Genie optimization software (364). In addition to general engineering optimization issues, the present method adds the following issues to the development of a global model as presented in this paper:

• Coordination of input and output. This requirement forces the development to be coordi­nated by only a few people and therefore presents the most significant “bottleneck" for in­dustrial application Currently we have one input coordinator for each geographic location. Changes to the model arc only made based on consensus between these representatives.

• What to model? In principle we are always modelling cost and benefit. Clear paths should be established between the major design parameters and the cost equations The benefits of a noise suppressor are easily defined, the number of dB suppressed at a reference condition can simply be used as an input parameter But at what cost? Here is where the specialist is inval­uable. He will calculate how much benefit to the aircraft he can guarantee as a function of various physical quantities (jet exhaust velocity…) for what cost (thrust-loss and suppressor complexity..). An accurate prediction of these relations is of course to his professional advan­tage since that will indirectly determine his task in the intermediate detailed design

• Generality versus accuracy. Computers arc bad with rules of thumb since they do not know what is behind them, but they arc alsu fast and can solve the general equations upon which ihc rules of thumb arc based. So therefore it is helpful to base the model on physical realities, and not to fudge it to fit one existing data set.

• Consistency. It is desirable to use one model to determine one parameter. If this is not pos­sible make sure that each equation representing the same parameter contains the identical set of variables and that the value and the derivatives at the cross-over points are continuous if possible.

The present method has a number of advantages over more traditional design methods.

This nietlwd.

• clarifies the goals of a design project and provide a means of communication between the disciplines. Most of the time it is not very clear what the objective of an aircraft design is By – agreeing on an objective function (for instance TOC) and a number of constraints it is possi­ble to settle interdisciplinary differences.

• automatically debugs the analysis routines. Small model inconsistencies are usually not no­ticed by the expert user because he only trusts his model in a limited range. The optimizer will exploit any weakness in the model and therefore make it more visible.

• cleanly compares between competing configurations. To compare aircraft they have to be analyzed with the same technologies and missions. In addition they have to be preferably an­alyzed w ith the same set of equations

• reduces the number of detail-design cycles. As the experiences at DA have shown, a good baseline design will cut the number of follow – on detail cycles, thus significantly reducing the total required time. •
sign space of best configurations. Active and nearly active constraints can be monitored to indicate important performance criteria or important technologies.

• allows the progress that is made in the analysis routines to be directly translated into an improved design. This provides a great deal of motivation to the individual who makes a contribution to an improvement in the model.

Some of the advantages of the current system are also described in a paper by Reimcrs

1366].