Design analysis of composite structures generally requires the use of computer software. Stiffness matrices, stresses and strains must be calculated for each ply and for the laminate as a whole. Failure criteria are calculated on a per-ply basis. Analytical programs based on laminate theory can be used for initial sizing of simple structures such as beams, rectangular plates and cylindrical shells. Finite element analysis is required for all but the simplest geometries and loading conditions.
Many programs have optimization routines, but these work primarily on geometry. A few can optimize layups (with significant restrictions), and even fewer can optimize for materials. Thus, it is important to narrow down the material selections early in the design process, and to have a rough idea of what layups will work.
Because of the inherent uncertainties in the analysis, and because composites are relatively new materials, most projects require some sort of structural testing. Building a full-scale prototype requires expensive tooling, so testing usually begins with subscale structural models to verify the analytical models and material property data.
Once a design has been settled upon, a dedicated test prototype is built, usually on the production tooling. Any design changes from this point on should be minor, requiring no change in the tooling.
The prototype is tested under loads that simulate the field conditions as closely as possible. Models are matched against the test data, which may include strain, displacement, acceleration, temperature or other important parameters. Any discrepancies between the models and the test data must be resolved before production can begin.
Testing usually continues throughout the production process. If a limited number of parts is being built, each one may be tested to representative acceptance loads. For large production runs, usually only a few parts will be tested. Production tests should include quality control checks, such as resin content and fiber orientation.