Oil-free centrifugal chiller optimal operation

Centrifugal chillers, using variable-speed turbo compressors with magnetic bearings, are becoming very common in Heating, Ventilation and Air Conditioning (HVAC) systems. Such solution guarantees superior energy efficiency, mostly under part load conditions, compared with traditional equipments, and it provides additional advantages such as light weight and a compact package. On the other hand, turbo machinery adds its own complexity to the whole HVAC system and its efficient management is a non-trivial task. In this paper a hybrid optimisation technique is employed to determine optimal operation, under various working conditions, for air-condensed water centrifugal chillers. The proposed method provides optimal solutions using a combination of two algorithms: A random population-based optimiser, the Gravitational Search Algorithm (GSA), followed by the deterministic Levenberg-Marquardt (LM) algorithm. The hybrid method effectively overcomes the problem of high sensitivity to initial conditions of LM technique and a shortcoming of GSA which reduces its searching efficiency when close to the optimum. The hybrid method has been tested in a Matlab^®-based simulation environment where the performance of an air-condensed centrifugal chiller is adequately evaluated. Simulation results guarantee high energy efficiency in a wide range of chiller working conditions.