Basically, you will have to understand that the thermal heat transfer from any device is through the 3 modes - Conduction, Convection and Radiation.
Hyperlynx can address all these aspects. As a generic assumption you should note that the analysis methodology in Hyperlynx is using Finite Differences method and hence the computation is faster and the accuracy of the results would be about +/-5% from actual measured values provided the input data is correct.
In Hyperlynx thermal, the data requirements are : power dissipation of all the components on the PCB, Component library build - this defines the type of component like SMT, Through-hole, etc, then thermal resistance of the component and its maximum junction and case temperature. The next thing is to define the environment values that would define how is the PCB mounted - single PCB, PCB in a rack - should mention if it is in the end slots of the rack or any of the intermediate slots, the air flow, how is PCB mounted - horizontal /vertical and air flow. The data validation takes some time. If the company follows a standard part naming convention like in IPC 7351, then these part libraries can be reused. One aspect to be noted is that the thermal parameters are package specific - that means any component that uses a PQFP100 footprint would have same thermal resistance and physical parameters as they adhere to JEDEC specification. Only power dissipation would vary based on the device functionality.
Once the data is validated, running analysis is simple. If you are also analyzing for power integrity, then you can perform PI-Thermal co-simulation to take the current density impact in plane layers for thermal profiling.
Trust this addresses your concern.