Particle breakage can significantly change the fabric (size and shape of particles and contact network) of a granular material, affecting highly the material's macroscopic response. In this paper, oedometric compression tests are performed on zeolite specimens and x-ray computed micro-tomography is employed, to acquire high resolution 3D images of the specimens throughout the test. The images are processed, to describe breakage spatially and quantify it throughout the test and gain information about the mechanisms leading to particle breakage. In addition to the image processing, the discrete element method (DEM) is used to study the initiation and likelihood of particle breakage, by simulating the experimental test during the early stages of loading and using quantitative results from the images to inform and validate the DEM model. A discrete digital image correlation is used, in order to incrementally identify intact grains and simultaneously get results about the strain field within the specimen, as well as the kinematics of individual grains and fragments. In the initial stages of breakage, there is a clear boundary effect on the spatial distribution of breakage, as it is concentrated at the moving boundary (more than 90% of total breakage) and circumferentially (more than 70% of total breakage) close to the apparatus cell. The DEM model can reproduce the bulk response of the material until the point where substantial breakage governs the macroscopic response and it starts to soften. Additionally, there is an initial indication that the spatial distribution of the force network matches the localisation of breakage radially, but it does not seem to localise close to the loading platen. This analysis will enrich our understanding of the mechanisms and evolution of particle breakage.