3D city models are being increasingly adopted by organisations in order to serve application needs related to urban areas. In order to fulfil the different requirements of various applications, the concept of Level of Detail (LoD) has been incorporated in 3D city models specifications, such as CityGML. Therefore, datasets of different LoDs are being created for the same areas by several organisations for their own use cases. Meanwhile, as time progresses newer versions of existing 3D city models are being created by vendors. Nevertheless, the existing mechanisms for representating multi-LoD data has not been adopted by the users and there has been little effort on the implementation of a mechanism to store multiple revisions of a city model. This results in redundancy of information and the existence of multiple datasets inconsistent with each other. Alternatively, a representation of time or scale as additional dimensions to the three spatial ones has been proposed as a better way to store multiple versions of datasets while retaining information related to the corresponding features between datasets. In this paper, we propose a conceptual framework with initial considerations for the implementation of a 4D representation of two states of a 3D city model. This framework defines both the data structure of such an approach, as well as the methodology according to which two existing 3D city models can be compared, associated and stored with their correspondences in 4D. The methodology is defined as six individual steps that have to be undertaken, each with its own individual requirements and goals that have to be challenged. We, also, provide some examples and considerations for the way those steps can be implemented.