Numerical method to select chiller sequencing control concerning uncertainties

Chiller sequencing control is used in multiple-chiller plants to switch on/off chillers according to building cooling load, aiming to achieve high energy-efficient while fulfilling the indoor thermal comfort demand. Various sequencing controls have been developed and implemented in practice, and typical ones include: total cooling load-based sequencing control, return water temperature-based sequencing control, direct power-based sequencing control, and bypass flow-based sequencing control. Each control uses a direct or an indirect indicator to represent building instantaneous cooling load and compares its indicator with pre-defined thresholds to determine sequencing action. Since different controls use different load indicators and thresholds, they suffer from different types of uncertainties and show different robustness to uncertainties. To select the most suitable control for a given chiller plant, it is necessary to evaluate the performance of control alternatives and their robustness when subject to uncertainties. To this end, in-situ tests or detailed simulation should be carried out. However, in-situ tests may be time consuming and cost inefficient; while detailed simulation needs complex models of the chiller plant, which is knowledge/effort-intensive. To simplify the selection, a numerical method is proposed based on the fact that the sequencing is only determined by the load indicators and the switch-on/off thresholds and all the potential uncertainties can be shifted to the load indicators. Details of the numerical method will be given. Three performance indices, including chiller total switch number, under-cooling percentage and energy use, are used to quantify the performance and robustness of the typical controls. A multi-criterion decision making method is adopted to select the most suitable control. Case studies show that the proposed method can assess the performance of the candidate control strategies efficiently without any complex models of the chiller plant.