ESA's Gaia mission aims to create a complete and highly accurate stereoscopic map of the Milky Way. The stellar parallaxes will be determined at the micro-arcsecond level, as a consequence the measurement of the stellar image location on the CCD must be highly accurate. The solar wind protons will create charge traps in the CCDs of Gaia, which will induce large charge loss and distort the stellar images causing a degradation of the location measurement accuracy. Accurate modelling of the stellar image distortion induced by radiation is required to mitigate these effects. We assess the capability of a fast physical analytical model of radiation damage effects called the charge distortion model (CDM) to reproduce experimental data. To realize this assessment we developed a rigorous procedure that compares at the sub-pixel level the model outcomes to damaged images extracted from the experimental tests. We show that CDM can reproduce accurately up to a certain level the test data acquired on a highly irradiated device operated in time delay integration mode for different signal levels and different illumination histories. We discuss the potential internal and external factors that contributed to limit the agreement between the data and the charge distortion model. To investigate these limiting factors further, we plan to apply our comparison procedure on a synthetic dataset generated through detailed Monte-Carlo simulations at the CCD electrode level.