A chiller plant consists of chiller, cooling tower, and pump subsystems. Two major configurations, primary-only and primary–secondary systems, are often used. Given the high energy costs of a plant, chiller plant operation optimization is important to save energy. For both configurations, chilled/condenser water supply temperatures are critical in improving chiller efficiency and should be considered as decision variables. However, nonlinearity of the problem is increased since chiller power consumption is a highly nonlinear function of these temperatures. Additionally, the problem is combinatorial considering the number of active units (e.g., chillers). In this paper, primary-only systems with identical units in each subsystem and primary–secondary systems with units of two sizes are studied, and both supply temperatures are optimized for energy savings. To obtain near-optimal solutions efficiently, a recent decomposition and coordination approach with little multiplier zigzagging and fast reduction of coupling constraint violations combining with sequential quadratic programming (SQP) is used. Penalties for the constraints that are difficult to be satisfied (e.g., mass balance constraints between fixed-speed pumps and variable-speed chillers) are added. After decomposition, complexity and nonlinearity of a subproblem are reduced drastically as compared with the original problem so that SQP is used. Numerical testing demonstrates that our approach is efficient in obtaining near-optimal solutions, and major energy savings are achieved as compared with benchmark strategies. The approach is scalable and can be used for chiller plant optimization and beyond. Note to Practitioners —Traditionally, chiller plant operation is often based on rules. For example, the number of active chillers is the minimum number that satisfies cooling requirements, and chilled water supply temperature is constant. Chiller power consumption is a nonlinear function of chiller cooling load and chilled/condenser water supply temperatures. Energy is wasted since chiller efficiency is low with fixed supply temperatures, and sometimes energy consumption of two chillers is less than that of one chiller. To save energy, chiller plant optimization with good decision variables such as the number of active units and chilled/condenser water supply temperatures is studied. The problem is challenging with high nonlinearity caused by considering such temperatures as decision variables. Furthermore, with discrete variables (e.g., the number of active chillers), the problem is combinatorial. To efficiently solve the problem for high-quality solutions, a novel decomposition and coordination approach is developed. Complexity and nonlinearity of a subproblem are reduced drastically after decomposition as compared with the original problem so that appropriate nonlinear methods are used to solve the subproblems. The results show that the solutions are near-optimal with short computational time and the approach is scalable. Additionally, major energy savings are achieved as compared with benchmark strategies. The approach provides a new and powerful way to solve chiller plant optimization problems and beyond.
Read full abstract