Demand estimation has long been an important topic that has recently received increased interest due to the development of real-time network modeling. For implementation to real networks, demand estimation is challenged by limited indirect hydraulic measurements (e.g., flow). Also, since many real network models have 104-105 demand nodes but only have hydraulic measurements in the range of 101-102, the demand estimation problem must be reduced to ensure that the problem is not under-determined. One approach to reduce the scale of the problem has been to generate clusters of nodes such that the demands for each cluster, rather than each individual node, can be estimated. Our recent research has shown that the resulting demand estimation problem can be sensitive to the availability and location of observed measurements as well as the clusters generated. Thus, the objective of this study is to provide a more in-depth understanding of the relationship between these two components in order to further develop real-time demand modeling capabilities. The research will evaluate the impacts of measurement locations and cluster generation on the demand estimation problem through more in-depth evaluations using the theoretical foundations of non-linear parameter estimation approaches to assess the identifiability of the resulting demand estimation problem. The expectation of this research will be a deeper understanding of the relationship between measurement locations and cluster approaches to generate adequate demand estimates. The insight gained is expected to improve our ability to generate cluster demands and provide improved metrics for placing additional measurement locations.