A DFT calculation can't replace an actual thermodynamic measurement but you can use DFT+statistical thermodynamics to get a glimpse. Using statistical thermodynamics, you can calculate free energies as a function of temperature and therefore maybe predict the phase transition.

Please note, however, that this is often done in harmonic approximation which means that if you investigate temperatures in which anharmonicities (=bond breaking) become important, your simulation values will be off.

Thanks Jürgen Weippert for your reply. Let me explain the problem a bit more. Two different studies, one experimentally and one numerically (thermodynamic calculations) have proved presence of a phase within liquidus projection but they are in conflict on composition range. e.g. one says Me-Si(24-25 at.%) is the primary solidification range and the other says e.g. Me-Si(22-27 at.%) is the primary solidification region of the referred phase. Meaning, when a molt with the nominated compositions freeze, the first arising phase would be the intermetallic phase. My idea was to check the energy of possible phases located in the vicinity of the intermetallic phase with considering a constant composition and temperature to see which phase posses the lower potential, and therefore more probability of formation. I don't know how realistic or possible is it.

Phase stability for bulk materials can also be found by the enthalpy of formation w.r.t. material's competing phases. For more details please read the following articles

10.1088/1361-648X/ab2bd1, 10.1103/PhysRevB.81.024111, 10.1038/s41598-021-82402-1.

Thanks Muhammad Waqas Qureshi

I found the provided papers very useful!