Hierarchical Systems Level Thermal Management for Multiple High Transient Heat Loads

With the proliferation of high power photonic devices and high density electronics packaging, there is a significant need for thermal management solutions at the systems level. This need is most acute for large transient loads, as the perfor- mance of traditional steady state designs may be unsatisfactory or unacceptable. The vapor compression cycle (VCC) is a well known cooling technology suitable for high heat dissipation requirements. It may be used to cool multiple heat sources using the same main refrigeration loop, increasing the overall efficiency of the system. However, designing and controlling a VCC for multiple heat loads operating under varying conditions remains challenging. Existing approaches are mostly based on open loop control for static operations. Such approaches are susceptible to instability due to the incursion of the critical heat flux (CHF), coupling between multiple interconnected evaporators, and transition between widely separated operating conditions. As a result, existing control designs are typically conservative, operating at inefficient low exit flow quality for the evap- orators. This chapter presents an alternative method to systems level thermal management that directly takes the system dynamics into account. This model- based hierarchical approach consists of three stages: static open loop control, robust feedback control, and predictive feedforward control. The static control sets the control inputs (valve openings, compressor speed, and accumulator heat input) at constant values for a set of constant heat loads. Design criteria involves maximizing the coefficient of performance (COP) while guaranteeing a desired CHF margin. The feedback controller uses sensor measurements to adjust the control inputs to reduce the effect of load variations on evaporator exit wall tem- perature at each operating point. The design needs to be “robust” to account for the model uncertainties, and cover the entire operating envelope. The pre- dictive feedforward control prepares the system for anticipated upcoming heat loads, and, in the case of multiple evaporators, distributes flows between loads to handle unexpected demands. This hierarchical control methodology of combining static, feedback, and feedforward controllers has been successfully demonstrated in nonlinear simulations and in a three-evaporator VCC testbed.