if the material is brittle in nature (with ..PP) the the basic triangular law can be use. in loose sense it reflects the smeared cracking approach. Gf=0.5 * sigma_f * strain_f * characteristic length. the failure values can be obtained from the literature for your fiber vol. fraction. it gives you approx Gf.

The amount of factors affecting this are large, here's a non-exhaustive list; type of glass fibre, fibre sizing, fabric type (UD, NCF, woven...?), processing method (a wet layup part will have a very different VF, porosity, fibre misorientation, etc...to an RTMed part, for example), resin modulus, resin (true) yield strength, resin strain to failure, fibre packing arrangement/interfibre distance, cure cycle, part thickness (this probably won't be a issue with a slow cure or most prepreg systems), etc... if you're using a well known prepreg system then you might be able to get much of the data you need for something like a Hashin damage model from the literature...if not then you'll need to test the materials that will be used in the part you are modelling using panels for test specimens that are made using the same process and of similar dimensions/scale as the part you are modelling (or look into the work on scaling up of composites processing by Anoush Poursartip and Goran Fernlund for guidance on issues that might arise on scale up on part quality/material defects)

Thank you very much for the answer

if the material is brittle in nature (with ..PP) the the basic triangular law can be use. in loose sense it reflects the smeared cracking approach. Gf=0.5 * sigma_f * strain_f * characteristic length. the failure values can be obtained from the literature for your fiber vol. fraction. it gives you approx Gf.