To better understand eruptive events in the solar corona, we combine sequences of multi-wavelength observations and modelling of the coronal magnetic field of NOAA AR 8210, a highly flare-productive active region. From the photosphere to the corona, the observations give us information about the motion of magnetic elements (photospheric magnetograms), the location of flares (e.g., H$\alpha$, EUV or soft X-ray brightenings), and the type of events (H$\alpha$ blueshift events). Assuming that the evolution of the coronal magnetic field above an active region can be described by successive equilibria, we follow in time the magnetic changes of the 3D nonlinear force-free (nlff) fields reconstructed from a time series of photospheric vector magnetograms. We apply this method to AR 8210 observed on May 1, 1998 between 17:00 UT and 21:40 UT. We identify two types of horizontal photospheric motions that can drive an eruption: a clockwise rotation of the sunspot, and a fast motion of an emerging polarity. The reconstructed nlff coronal fields give us a scenario of the confined flares observed in AR 8210: the slow sunspot rotation enables the occurence of flare by a reconnection process close to a separatrix surface whereas the fast motion is associated with small-scale reconnections but no detectable flaring activity. We also study the injection rates of magnetic energy, Poynting flux and relative magnetic helicity through the photosphere and into the corona. The injection of magnetic energy by transverse photospheric motions is found to be correlated with the storage of energy in the corona and then the release by flaring activity. The magnetic helicity derived from the magnetic field and the vector potential of the nlff configuration is computed in the coronal volume. The magnetic helicity evolution shows that AR 8210 is dominated by the mutual helicity between the closed and potential fields and not by the self helicity of the closed field which characterizes the twist of confined flux bundles. We conclude that for AR 8210 the complex topology is a more important factor than the twist in the eruption process.