Why Do Design Optimization?

Chapter from: "Why Do Design Optimization?" by D.Spicer

Optimization promotes automation

It is generally recognized that the direct use of the main design data model as a unifiying agent in the design process forces it to become more seamless and integrated. This in turn tends to force clarification of the process itself, and a reduction in cycle time. In order to carry out any significant design optimization work, an automated design evaluation capability is a prerequisite. A move towards optimization therefore reinforces this beneficial trend.

It may be though that the direct use of the design data model for optimization is problematic, because currently CAD data transfer is rather slow. However, in most optimizations the amount of data being moved may well not be so large as to dominate the calculation, compared with evaluation. There is also an aspect which is an advantage. In traditional CAD work, any of the problems associated with CAD data transfer arise through transferring from one CAD system to a different one. But in the data transfer needed in optimization we are transferring from the CAD system to itself. So many of the problems and fixes, such as having to remove CAD elements which have been found to give compatibility problems, are not relevant in optimization.

Optimization promotes focus on metrics

It is argued that the clarification of the design’s measures of success is essential to ensuring that the design team are addressing the right set of problems for the business. The use of optimization ensures that the issue is brought clearly into focus and that the sometimes very difficult task of ensuring that sensitive areas of the problem receive their full attention.

Optimization promotes closing gaps in design evaluation toolset

The use of top-down design metrics implies a completeness of capability to evaluate there metrics. Not all the areas of design are currently free from difficulties in this respect.

Analysis challenges:

Many engineering analysis problems exist, for example in fluid dynamics and electromagnetics, for which only a limited solution capability exists due to their needs for amounts of computing power and resource which are not available for everyday needs.

Modeling challenges:

Various areas of design exist in which there are still significant challenges in developing clear agreed models of the design attributes concerned; example areas are costing and systems modelling

Data challenges

The complete data model of a major engineering product such as an aircraft or a ship is currently much too large to be convenient for participating in “whole design” computations.

The use of optimization enforces a need to provide design evaluation tools in all the areas of design which are key to the important design choices.

The requirement is not to make large leaps in technology overnight, but to fairly capture the ways in which the design is evaluated in practice by the design team and formally acknowledge these.

Optimization promotes design for analysis

To automate idealization, and provide closer integration between the Design office and the analysis departments (Aerodynamics, Structures, Electromagnetics etc) the activities of:

• Definition of the outside of the product
• Geometry repair
• Finite element model preparation
• Elimination of detail

have to be closely considered as part of the design process, rather than as unscheduled labor intensive and time-consuming procedures. It is typical for design data to be created to a standard which is unfit for any of the above operations. Without automation of these, full use of CAD models cannot be achieved, and even the use of analysis models will be less effective. The use of optimization will undoubtedly bring these automation issues into focus.

Optimization promotes Definition and use of Operating Scenarios

Top level design metrics must inevitably reflect how the product is manufactured, operated in the field, and supported. The production and operating scenarios provide a framework to bring into focus the important cost, performance and customer satisfaction issues. The scenarios to be defined may need to encompass any or all of the following areas:

1. Strategic studies affecting procurement
2. Operational analysis of in-service use
3. Manufacturing cost models
4. Business models: (single company site, dispersed company, collaborative project)

Optimization provides a means of rapid organizational learning

Often, an organization’s know how is what keeps it in business. This know how is the collective distillation of experience from having done similar jobs before and build up knowledge of the designs. Design optimization, which allows thousands of designs to be experimented with on a computer, speeds up learning process. By trying out designs which are diverse, it promotes a broadening of knowledge of the solution possibilities and increases engineering understanding. If the results of investigating a broader design range are condensed and stored suitably, an advanced knowledge base can be built for future reference.

Optimization provides the means of outperforming the competition

For many industries, time-to-market is a key issue. The ability to consider and evaluate many designs in a short period of time may confer a decisive advantage over the competition.