Machine to machine communications (M2M) are retained as one of the most important enabling technologies for the fulfilment of the Smart City paradigm. Unfortunately, M2M communications are still far to be a fully viable technology. One of the most important research topics concerning M2M is how to extend, as long as possible, the lifetime of the devices involved in the communication process by: (i) reducing the energy consumption and, (ii) using an harvesting system as an optional capability to recharge the devices. However, one important issue hindering the research advancement of this topic is the lack of an accurate simulation environment. Although the network simulator-3 (ns-3) offers many features to resolve that issue, the natively embedded models of energy harvesting systems are very simple and fail to take into account many aspects of the harvesting process. In this paper, we focus on solar energy harvesting systems and we propose an extension to ns-3 to accurately model them. Through simulation results, we show how our model allows to realistically simulate the behaviour of the devices powered by solar energy harvesters with different time scales ranging from a few seconds to several days.