Navigation, also known as discovering one’s direction, is a complex human activity. To produce effective routes, it relies on knowledge of the surroundings’ precise geometry and semantic information. Complex geometrical data can be precisely delineated with the improvement of 3D geometric models. A precise 3D geometric model containing a specifically built-in Building Information Modelling (BIM) environment can be integrated into the Geographical Information System platform for indoor path generation to satisfy the requirements of indoor location-based services. Therefore, this paper proposes an approach to evaluate a 3D indoor topology network called a lattice topology data model (LTDM) for the floor-level paths in a 3D multipatch-based model. The LTDM requires the geometric information of the integrated BIM model to identify the indoor space and bounding lines for indoor network generation. The novelty of this study is in the application of the replacement of cell values into vector length for pathfinding through a combination of the Poincaré duality theorem and Dijkstra’s algorithm. The Campus Infrastructure Building model was chosen to validate the proposed method. Multiple space centroid pairs within the floor level were randomly selected to identify the shortest path using the LTDM principle. Paths drawn from the Medial Axis Transformation were compared with LTDM-generated paths for availability testing. The average floor-level path availability was 112% due to the generation of extra paths reflecting real-life situations. The LTDM paths were compared with on-site measurements for accuracy tests, and the average error rate was 3.18%. The results show that the implementation of the LTDM generates an excellent topology data network.