Chiller Sequencing Control under Uncertainties


Student thesis: Doctoral Thesis

View graph of relations


  • Yundan LIAO


Awarding Institution
Award date25 Jul 2016


Chiller sequencing control is an essential function for multiple-chiller plants that automatically switches chillers on and off in terms of building instantaneous cooling load, aiming at achieving an overall energy efficiency while fulfilling the cooling demand. Its control performance significantly affects the reliability of cooling supply, the stability of chiller plants and the efficiency of building energy. However, due to the uncertainties in operating conditions, chiller sequencing control cannot operate as reliably as anticipated and usually fail to achieve the expected control performance. Therefore, this thesis provides a systematic study on chiller sequencing control under uncertainties and several methods to enhance the control performance and robustness of chiller sequencing control under uncertainties.
In order to systematically study the uncertainties associated with chiller sequencing control, uncertainty analysis and robustness analysis are conducted. Uncertainty analysis aims to investigate the impacts of uncertainties on the performance of chiller sequencing control. To this end, the uncertainties in chiller sequencing control are firstly identified and categorized. Then, an uncertainty shifting method is developed to facilitate uncertainty propagation in computer simulations. Finally, the impacts and significances of uncertainties are assessed with simulation studies.
Robustness analysis aims to quantify the robustness of chiller sequencing control, where four typical control strategies are considered, including total cooling load-based sequencing control, chilled water return temperature-based sequencing control, bypass flow-based sequencing control and direct power-based sequencing control. The robustness of the four control strategies is evaluated according to the variations of predefined performance indices when they are subject to multiple levels of uncertainties. Results of the robustness analysis are used to demonstrate that different control strategy have different sensitivity to different type of uncertainties.
Aided by uncertainty analysis and robustness analysis, a numerical method is developed to select a most suitable control strategy for a given chiller plant under uncertainty instead of proceeding this selection via detailed simulation or tedious in-situ tests. In this numerical method, the multiple-chiller plant model is simplified with a regression method. The necessary information for robustness analysis is obtained from historical operation data, and Monte Carlo method is adopted to simulate the stochastic characteristic of uncertainties. A combined index is provided to incorporate the outcomes of multiple criteria and the preferences of decision makers into the selecting process. Case studies demonstrate that this numerical method can provide reliable results as detailed simulations but it provides much convenience in selecting a suitable control strategy according to the chiller plant operating conditions.
To enhance the robustness of chiller sequencing control, two methods are proposed according to the characteristics of control strategies under uncertainties. The first one is to enhance the direct power-based control by using the chilled water supply temperature as a subsidiary criterion for switch-on control. The second one is to enhance the robustness of the chilled water return temperature-based control and the bypass flow-based control by hybridizing them with the direct power-based control. The rationality of the enhanced controls is validated by site operation data analysis. The performance of them is evaluated via robustness analysis and comparisons with conventional controls. Robustness analysis shows that the enhanced controls have significantly improved performance under uncertainties and the comparison studies demonstrate that the enhanced controls can provide more reliable performance than the conventional controls. Such good robustness and reliable control performance indicate that the enhanced controls are suitable for practical applications when multiple uncertainties exist.