MINLP‐based hybrid strategy for operating mode selection of TES‐backed‐up refrigeration systems

Abstract

This brief deals with the satisfaction of the daily cooling demand by a hybrid system that consists of a vapor-compression refrigeration cycle and a thermal energy storage (TES) unit, based on phase change materials. The addition of the TES tank to the original refrigeration plant allows to schedule the cooling production regardless of the instantaneous demand, given that the TES tank can store cold energy and release it whenever deemed appropriate. The scheduling problem is posed as an optimization problem based on mixed-integer nonlinear programming (MINLP) since it includes both discrete and continuous variables. The latter corresponds to the references on the main cooling powers involved in the problem (cooling production at the evaporator and TES charging/discharging), whereas the discrete variables define the operating mode scheduling. Therefore, in addition to the hybrid features of the physical plant, a hybrid optimal control strategy is also proposed. A receding horizon approach is applied, similar to model predictive control (MPC) strategies, while economic criteria are imposed in the objective function, as well as feasibility issues. The TES state estimation is also addressed since its instantaneous charge ratio is not measurable. The proposed strategy is applied in simulation to a challenging cooling demand profile, and the main advantages of the MINLP-based strategy over a nonlinear MPC-based scheduling strategy previously developed are highlighted, regarding operating cost, ease of tuning, and ability to adapt to cooling demand variations. KEYWORDS mixed-integer nonlinear programming, phase change materials, refrigeration system, scheduling, thermal energy storage 1 INTRODUCTION Refrigeration cycles based on vapor compression constitute the worldwide leading technology for cooling issues, ie, air conditioning, medium-temperature refrigeration, and freezing. Very different areas demand controlling room temperature, for instance, for human comfort, food storage and transportation, industrial processes, etc, where a wide power range is involved, from less than 1 kW to above 1 MW. 1 It is stated that about 30% of the total energy all over the world Abbreviations: MINLP, mixed-integer nonlinear programming; TES, thermal energy storage; MPC, model predictive control.

Distributed iterative learning temperature control for large‐scale buildings

“…Another typical strategy is the model predictive control (MPC) 8‐11 . Based on the system model, MPC can predict system states and generate an input signal to optimize a control index 9 .…”

“…Another typical strategy is the model predictive control (MPC). [8][9][10][11] Based on the system model, MPC can predict system states and generate an input signal to optimize a control index. 9 However, the accurate system model is hard to find in practice due to the complexity of building materials, stored items, and human activities.…”