In the conventional approach, the material properties of laminate composites for crash simulations are typically obtained from standard coupon tests, where the test results only provide single layer material properties. However, the lay-up effects for the failure behaviors of the real structure were not considered in numerical simulations. Hence, there was discrepancy between the crash simulations and experimental tests. Consequently, an intermediate stage is required for accurate predictions. Some component tests are required to calibrate the material models in the intermediate stage. In this paper, a laminate cylinder tube under high-impact velocity in the direction of tube axis is chosen as an example for the crash analysis. The tube consists of 24 layers of uni-directional (UD) carbon fiber composite materials, in which 4 layers are perpendicular to, while the other layers are parallel to the impact direction. An LS-DYNA meso-model was constructed with orthotropic material models counting for the single-layer material behaviors. Between layers, a node-based tie-break contact was used for modeling the delamination of the composite material. Since fiber directions are not single-oriented, the lay-up effects could be an important effect. From the first numerical trial, premature material failure occurred due to the use of material parameters obtained directly from the coupon tests. Some parametric studies were conducted to identify the cause of the numerical instability. The finding is that the material failure strength used in the numerical model needs to be enlarged to stabilize the numerical model. A hypothesis was made to provide the foundation for enlarging the failure strength and the corresponding experiments will be conducted to validate the hypothesis. The hypothesis is about the confinement effect of the resin materials. To prove the hypothesis, RVE analyses of UD CF composites were conducted, where the material strength could be enlarged 2.7 time when material was constrained.