The goal of online optimization is to find economizing input values that minimize the plant's steady-state operational cost. Often, the optimal value of these inputs is a function of the system's disturbances. For systems with a direct measurement of the operational cost function and multiple process outputs, there is a wealth of information that can be exploited for online optimizing feedback control. When extremum seeking control is applied to these systems, multiple potential process measurements unrelated to achieving system performance objectives present a choice of the extremum seeking controlled variable. In this paper, a Vapor Compression System moving boundary simulation model is employed to investigate the effectiveness of combining self-optimizing control with extremum seeking control. Results show that combining extremum seeking's ability to adapt to slowly varying disturbances under minimal assumptions about the system model with the transient performance guarantees provided by self-optimizing control improves optimization performance by nearly 60% relative to the case where extremum seeking directly controls an actuator input.