Laminate Statistical Allowable Generation for Fiber-Reinforced Composite Materials: Lamina Variability Method

摘要

Substantiation of composite structures often requires analysis as well as tests conducted at multiple building-block levels. Typically, as one proceeds up the building block and becomes more confident with the analytical design methods, the number of samples or replicates is reduced since the structural complexity increases. As the complexity of higher-level tests increases, more variables are introduced; i.e., fastener holes and multidirectional laminates. Currently, there is no standardized statistical methodology that allows progression up the pyramid of tests with a concurrent reduction of the number of tests while maintaining statistical reliability established at the lower replicate level. Therefore, conservative approaches are exercised at the higher level of building-block tests to be able to use the material variability of lower levels to represent the higher levels. Because a reduction in the number of samples results in excessively low or overly conservative material allowables, this research program explores the use of lamina-level variability to generate B-basis allowables for fiber-reinforced laminated composites. A multibatch laminate database was created to form baseline B-basis values. Then, B-basis allowables were obtained from single-batch data sets and compared with multibatch data to investigate the statistical reliability of the proposed small-sample B-basis methodthe lamina variability method (LVM). This investigation included six material systems from three prepreg manufacturers: Toray Composites, FiberCote Industries, and Advanced Composites Group. In addition to the proposed methodology, two additional small-sample, B-basis methods proposed by Steve Ward were compared. Analysis of lamina and laminate test data for six different prepreg materials showed that LVM works well and gives conservative B-basis allowables that are about 90% of the three-batch laminate allowables. Caution should be exercised if the laminate data exhibits unusual variability due to test methods. If the variability is due to the material, the LVM method should be used with caution. If test variability is too high, for example, bearing strength, then three batches of laminate tests should be performed for that property to obtain B-basis allowables. It is imperative to realize the advantages as well as the risks associated with small-sample methods, and the data may need to be validated with additional tests based on specific applications.