The properties of L-mode confinement have been investigated with a set of dedicated experiments in ASDEX Upgrade and with a related modelling activity with the transport code ASTRA and the quasi-linear turbulent transport model TGLF– SAT2, with boundary conditions at the separatrix. The values at the boundary have been set by the two-point model for the electron temperature, with the ion temperature proportional to the electron temperature by a constant factor, and the electron density set by a constant fraction of the volume averaged density. The influx of neutrals has been set through a feedback procedure which ensures that in the simulation the same particle content as in the experiment is obtained. The sensitivity of the results under considerable variations in the choice of the boundary conditions has been investigated and found to be limited. The predictions of this full-radius modelling set-up have been compared to experimental results covering a scan in electron cyclotron resonance heating (ECRH) power in both hydrogen and deuterium plasmas, a plasma current scan with fixed magnetic field, under both ECRH and neutral bean injection (NBI) heating, an increase in plasma density with costant ECRH power in hydrogen plasmas, as well as variations of the fraction of electron and ion heating at approximately constant total heating power, as well as a change of main ion from deuterium to hydrogen. The ASTRA-TGLF predictions have been found to reproduce all of the experimentally explored dependencies with relatively good accuracy, providing evidence, for the first time to our knowledge, that the main properties of L-mode confinement can be reproduced by conventional full-radius transport modelling with a quasi-linear turbulent transport model. Evidences of largest disagreement, although usually not exceeding the 20%, have been found at high electron heating power, where TGLF underpredicts the electron and particularly the ion thermal stored energies, and in the current dependence of confinement, which, in electron heated conditions, is predicted to be weaker than in the experiment.