This article is primarily concerned with the optimal configuration of multi-energy microgrid in cold climate. Although involving photovoltaic (PV) has been commonplace in works of this sort, one thing that has yet to receive sufficient care is pertained to how exactly PV power output is modeled. Indeed, refined PV modeling has an decisive effect on whether or not the configured microgrid is going to operate as intended in the long run. On this account, physical model chain, a refined way to convert irradiance to PV power output through a series of models, as opposed to the conventional one-equation conversion, is herein emphasized. Then, by adopting the mixed integer linear programming, the configuration of a microgrid with both residential and commercial buildings in cold climate, alongside a rich selection of energy converters and storage devices, is performed. The case study results suggest: (1) configurations based solely on the conventional PV model tend to exaggerate the power production from PV by 10.31%, causing a severe underestimation in total annualized cost by 8.26%; (2) thermoelectric coupling in cold climate favors the inclusion of combined heat and power (CHP) units, which occupy 17% of total capacity; and (3) when heavy carbon emission penalty is invoked, the optimal configuration requires one to double the CHP installation but only add a moderate amount of PV. To support open research, data and Python code used to generate the results are offered as supplementary materials.