The dynamic congestion pricing problem of a large-scale transportation network is characterized with multiple expensive-to-evaluate objectives without closed forms. A computationally efficient simulation-based optimization (SBO) method is proposed to solve the dynamic congestion pricing problem with tight computational budgets. This paper develops an innovative multiobjective optimization approach that integrates simulation based dynamic traffic assignment, surrogate models for optimizing computationally expensive objective functions, and DIRECT that is a deterministic global computing with modification to Lipschitzian optimization. The authors investigate the existence of an invariant macroscopic fundamental diagram (MFD) for a real-world transportation network, and validate simulated MFDs using fixed freeway traffic flow measurements and probe vehicle data. To demonstrate the SBO framework with an application to the vehicle mileage traveled (VMT) based pricing of a freeway sub-network, this paper utilizes a calibrated simulation-based dynamic traffic assignment model of morning peak to evaluate system performance corresponding to time-of-day toll charges. Optimal scenarios of different objectives of average travel time minimization, throughput maximization, and toll revenue maximum are compared. The combined overall desirability function is optimized and performs much larger desirability than single-objective problems. The optimal result is more balanced and without short board in different objective functions. The SBO framework established in this study can support future simultaneous analyses of transportation policies, planning applications and operational strategies.