Abstract Many models for prediction of longitudinal tensile failure of unidirectional (UD) composites have been developed in the last decades. These models require significant assumptions and simplifications, but their consequences for the predictions are often not clearly understood. This paper therefore presents a blind benchmark of seven different models applied to two virtual materials. Reliably capturing the localisation of stress concentrations was vital in predicting the effect of matrix stiffness and strength on composite failure strain and strength as well as fibre break and cluster development. Although the models have different assumptions regarding stress redistributions around fibre breaks, the 2-plet (clusters of two fibre breaks) development was similar. Distance-based criteria were shown to be inadequate for monitoring cluster development. The discussions provide detailed insight into how the model assumptions are linked to the differences in the predictions.

中文翻译：

---

两种虚拟单向复合材料的七个纵向拉伸破坏模型的盲标定

摘要 在过去的几十年里，已经开发了许多预测单向 (UD) 复合材料纵向拉伸破坏的模型。这些模型需要大量的假设和简化，但它们对预测的影响往往并不清楚。因此，本文提出了应用于两种虚拟材料的七个不同模型的盲标。可靠地捕捉应力集中的定位对于预测基体刚度和强度对复合材料失效应变和强度以及纤维断裂和簇发展的影响至关重要。尽管模型对围绕纤维断裂的应力重新分布有不同的假设，但 2-plet（两个纤维断裂的簇）的发展是相似的。事实证明，基于距离的标准不足以监测集群发展。讨论提供了有关模型假设如何与预测差异相关联的详细见解。