Destroying galaxies (or not) with AGN feedback

Supermassive black holes (SMBHs) are believed to reside at the centres of most galaxies. Observations suggest that the host galaxies are strongly affected by feedback produced by accreting SMBHs. Feedback in the form of ultra-fast outflows (UFOs), which are expected to interact with the interstellar medium (ISM), have been used to explain scaling relations between SMBHs and their host galaxies. Such relations suggest that the feedback and ISM must couple very weakly, however, it is not clear how this is achieved. In this thesis I provide observational tests to constrain UFO shock physics. I show that if UFO shocks cool via inverse Compton (IC) scattering, they should be observable in X-rays, but are not actually seen. The likely explanation for this is that the outflow is in a two-temperature, non-radiative regime. This implies that AGN outflows do not loose their kinetic energy to radiation and that an alternative energy loss mechanism is needed to explain the weak coupling required. I use high-resolution simulations to investigate an UFO impacting upon a turbulent ISM. Complex processes occur in the turbulent medium, such as the detachment of mass and energy flows, which are missed in a homogeneous medium. While the shocked UFO can escape through low density regions, high density clumps are resistant to feedback and can continue to have negative radial velocities. Energy losses in the multiphase ISM may provide an alternative to the IC radiative loss mechanism. Given the importance placed upon simulations in aiding our understanding of AGN feedback, I present a resolution study using a commonly employed sub-grid feedback prescription. I find that changes in resolution impact upon feedback efficiency, although not necessarily in a systematic way. I therefore suggest caution when analysing simulation results in order to ensure that numerical artefacts are not interpreted as physical phenomena.