If by "unbalanced network" you mean the transmission line's parameters are different for each phase, you cannot use Pi Section Line block.
Else if the transmission lines are symmetric and the rest of the system is unbalanced, then in general, both models are accurate, they just use different modeling approaches. One may be more accurate than another in specific applications. For instance, for EMT studies, one will most likely have higher fidelity results using the Distributed Parameter Line. Note that higher fidelity comes at a trade off with simulation speed. I suggest referring to the documentation for each and determine which is best suited for your application.
For documentation on Distributed Parameter line: https://www.mathworks.com/help/physmod/sps/powersys/ref/distributedparameterline.html?s_tid=srchtitle