The measurement principle of Global Navigation Satellite Systems (GNSS) is based on measuring the one-way travel time of the respective signals from satellite to receiver. Therefore, the accurate synchronization of the atomic clocks onboard the satellites are key to achieve highest accuracy for precise applications of GNSS. The up-to-date GNSS use different on-board atomic frequency standards with a wide range of stability. The Galileo satellites are equipped with 2 highly-accurate Passive Hydrogen Masers (PHMs) and 2 Rubidium clocks, whereas other GNSS use different types of Rubidium clocks and in some cases even only Caesium clocks. Ground stations are usually equipped with less accurate clocks for practical reasons, yet an increasing number of International GNSS Service (IGS) sites is being set up with PHMs as well (currently over 70 stations, which represents about 15 %). With these highly accurate clocks available on satellites and ground stations, a modelling of the satellite and receiver clock corrections becomes feasible. However, even for such clocks with excellent frequency stability the clock corrections are typically being estimated epoch- wise without constraints in state-of-the-art Precise Orbit Determination. This leads to a high degree of freedom and a weakening of the parameters in the estimation process. Clock modelling on the other hand leads to more robust estimates for high-sampled parameters such as tropospheric corrections, as well as improvements in the orbits themselves. The influence of satellite clock modelling on the performance of precise GNSS dynamic orbit determination was investigated for a number of analysis scenarios with the focus on Galileo satellites running on PHMs. Different deterministic clock models were tested by applying absolute constraints on selected clocks. The presentation will give an overview of the different scenarios analyzed in this study and show the improvement, which can be achieved using clock modelling.