Reliable thermal management is becoming increasingly critical in modern data centers as computing density, workload variability, and the adoption of liquid and hybrid cooling continue to grow. This lecture introduces Modelica-based model development for data center thermal management, with a focus on building physically interpretable and reusable models of air-side, liquid-side, and facility cooling systems. Using representative examples from hybrid-cooled data centers, the lecture will discuss system design, component modeling, and validation using measured data. It will also illustrate how Modelica models can be used to create virtual testbeds for evaluating cooling performance, investigating spatial and temporal workload variability, and developing advanced control and optimization strategies. The lecture aims to provide a practical foundation for using physics-based modeling to support reliable, energy-efficient, and scalable data center cooling system design and operation.