In situ tissue engineering exploits the process of cell infiltration and extracellular matrix production for so-called endogenous tissue restoration (ETR) in biodegradable synthetic vascular grafts. Challenges remain regarding aneurysm and thrombus formation. In order to understand the biological processes underlying ETR, we propose an in silico approach through a computational model predicting changes in graft composition, i.e. gradual replacement of synthetic scaffold material by collagen. Wall stress and (time-averaged) wall shear stress (TAWSS) are considered important biomechanical stimuli that influence extracellular matrix production. Therefore, we propose two methods for fluid-structure-growth (FSG1 and FSG2) modelling of ETR, combining a finite element model, evolving material properties through a growth and remodelling (G&R) algorithm, and the computation of the TAWSS. With FSG1, the TAWSS is estimated at every time point of G&R through an artificial neural network surrogate model of rigid wall computational fluid dynamic models. With FSG2, the TAWSS is calculated every 60 time steps (60 days) of G&R through a two-way coupled fluid-structure-interaction model. Our verification study shows good agreement between both approaches (see Fig. 1), showing the reliability of lower fidelity modelling in the context of ETR, specifically in the context of this carotid conduit case-study. The FSG1 model is calibrated to match predicted graft diameter and constitution over time with measurements from sheep carotid conduit experiments (see Fig. 1). As validation, model outcomes are compared to unseen experiments, where a different initial graft geometry was considered. Predictions for graft constitution after six months are good, with predictions falling within the mean ± standard deviation range of experimental measurements, while the predicted graft diameter is reliable only up to 2 months of G&R (see Fig. 1).