To improve modeling of orthotropic structural composites, a new failure criterion has been created using multiscale modeling. Repeating Unit Cells (RUCs) and Representative Volume Elements (RVE) are used to model the constituent parts of the composite. The RVE is then subjected to multi-axial state of stress and the first instance of failure in the RVE is detected and tagged as a point in the failure point cloud data (FPCD) in the stress/strain space. To ensure that the data is sufficiently accurate, the generated data is validated against uniaxial laboratory test results before multi-axial states of stress via surface tractions and displacement-controlled loadings are applied on the RVE. The generated FPCD is then used as input in a commercial explicit finite element program. During finite element (FE) analysis, at each time step in every element and at every stress Gauss point, the state of stress at that point is queried to check if the state of stress is inside or outside the failure surface using k-nearest neighbor (k-NN) classification concept. A simple mesh regularization scheme is implemented to account for varying element sizes. The developed framework is tested using a unidirectional composite (UDC) commonly used in the aerospace industry. Results indicate that the framework is promising and can be readily extended for almost any composite architecture for the following reasons: accuracy (predictions are close to actual experimental tests), versatility (can be applied to almost any composite architecture), and ease of use (requires minimal calibration).