Clusters of galaxies provide a unique opportunity to simultaneously study cosmology through low scatter scaling relations and the complex baryonic physics that occurs in cluster cores. As such it is of key importance to quantify the effects of the various physical processes that drive cluster evolution. In this thesis a sample of 30 clusters from the Millennium Gas Simulation, of masses 10^14/h Solar Masses < M200 < 10^15/h Solar Masses, were selected and run at a higher resolution using the re-simulation technique, using a modified version of Gadget-2, an N-body SPH code. Each cluster was run multiple times with increasing levels of sub-grid physics in order to separate the different effects that govern cluster evolution. The models implemented starting with non-radiative (NR), simulations then added cooling and star formation (CSF), supernova feedback (SFB) and AGN feedback model (AGN) respectively. In order to best match observations a study of supernova and AGN feedback parameters was conducted. The sample of clusters were also used to quantify the magnitude of biases created when observing clusters, in an attempt to classify the accuracy of these measurements of clusters. Additionally, the effects of the biases were also included in the estimation of the cluster mass using hydrostatic equilibrium. The best match to the observed gas, star and baryon fractions, scaling relations and gas profiles was found when powerful supernova feedback was included, which heats gas particles to 10^7K, and an AGN model whose heating temperature scales with the final virial temperature of the cluster, so that particles in a 10^14/h Solar masses and 10^15/h Solar Masses cluster are heated to 10^8K and 10^8.5 K respectively. Outside the core, this model successfully matches all the observed profiles and scaling relations excluding the spectrascopic-like temperature. The core region is simulated with come success, with pressures matching those observed but gas that is too cool and dense, resulting in an inability to reproduce the non cool core entropy profiles. Cold dense gas is more heavily weighted in the spectrascopic-like temperature, allowing significant contributions from gas in substructures and cold dense clumps of gas that are un-ascociated with any substructures and seems to be an artificial construct of SPH. When this gas is removed using the method outlines in \cite{Roncarelli2006}, temperatures outside the core match observations, but the core region is still too dense and cool. Clearly this core region requires more complex physics, possibly through implementation of an improved SPH code or more complex sub-grid physics such as that associated with the AGN feedback. The bias profiles also exhibit a similar sensitivity to the cool dense gas clumps, having a profound effect on the observed profiles and creating significant scatter in the mass estimated using hydrostatic equilibrium. Removing this cold dense gas using the Roncarelli method results in reduced biases and hydrostatic mass estimates closer to the true values. The resulting scaling relations and profiles including the effects of biases differ from those without the biases, but not significantly. It is clear that biases can affect the observed profiles and scaling relations, but this effect is minimised by excluding the coldest densest gas. As the choice of how much gas is removed is somewhat arbitrary, it is clear that further work in this field would require better SPH implementations that do not produce the erroneous dense gas clumps and the generation of mock observations using the simulated data.