With the growing importance of efficient energy management in educational institutions, this study employs the K-means clustering technique with the Soft-DTW metric to classify active and reactive power measurements from service transformers at the Federal University of Juiz de Fora (UFJF) campus. The primary goal is to identify typical load curve profiles according to the characteristics of the loads, thereby characterizing the nature of usage of the buildings served by these transformers. The resulting typical curves enable more realistic network studies, such as energy procurement, transformer sizing, photovoltaic plant allocation, and electric vehicle charger placement. This approach addresses the gap in existing research by providing a robust method for clustering load profiles that can be applied to other universities where load curves are not measured. Clustering the load curves makes it feasible to design low-cost measurement systems by sampling only a small subset of transformers with similar load characteristics. The results demonstrated the superior performance of Soft-DTW over Euclidean distance in aligning and grouping time series with varied profiles. This enhanced clustering accuracy is fundamental for effective load profiling, supporting informed decision-making in power system management and planning. In conclusion, the advanced clustering techniques applied in this research provide a deeper understanding of electrical load management, suggesting potential extensions to other facilities and broader factors influencing consumption patterns. Ultimately, this research contributes to optimizing power systems to meet better the demands of a dynamic and evolving energy landscape.