It is now widely accepted that heating processes play a fundamental role in galaxy clusters, struggling in an intricate but fascinating ‘dance’ with its antagonist, radiative cooling. Last-generation observations, especially X-ray, are giving us tiny hints about the notes of this endless ballet. Cavities, shocks, turbulence and wide absorption lines indicate that the central active nucleus is injecting a huge amount of energy in the intracluster medium. However, which is the real dominant engine of self-regulated heating? One of the models we propose is massive subrelativistic outflows, probably generated by a wind disc or just the result of the entrainment on kpc scale by the fast radio jet. Using a modified version of the adaptive mesh refinement code flash 3.2, we have explored several feedback mechanisms that self-regulate the mechanical power. Two are the best schemes that answer our primary question, that is, quenching cooling flow and at the same time preserving a cool core appearance for a long-term evolution (7 Gyr): one is more explosive (with efficiencies ∼ 5 × 10−3–10−2), triggered by central cooled gas, and the other is gentler, ignited by hot gas Bondi accretion (with ε= 0.1). These three-dimensional simulations show that the total energy injected is not the key aspect, but the results strongly depend on how energy is given to the intracluster medium. We follow the dynamics of the best models (temperature, density, surface brightness maps and profiles) and produce many observable predictions: buoyant bubbles, ripples, turbulence, iron abundance maps and hydrostatic equilibrium deviation. We present an in-depth discussion of the merits and flaws of all our models, with a critical eye towards observational concordance.