In recent years, the transmission network expansion planning (TNEP) problem has become increasingly complex. As this problem is a nonlinear and nonconvex optimization problem, researchers have traditionally focused on approximate models of power flows to solve the TNEP problem. Until recently, these approximations have produced results that are straightforward to adapt to the more complex problem. However, the power grid is evolving towards a state where the adaptations are no longer as easy (e.g., large amounts of limited control, renewable generation), necessitating new approaches. In this paper, we propose a discrepancy-bounded local search (DBLS) that encapsulates the complexity of power flow modeling in a black box that may be queried for information about the quality of a proposed expansion. This allows the development of an optimization algorithm that is decoupled from the details of the underlying power model. Case studies are presented to demonstrate cost differences in plans developed under different power flow models.