Micro-grid energy dispatch optimization and predictive control algorithms; A UC Irvine case study

Distributed power and energy resources are now being used to meet the combined electric power, heating, and cooling demands of many buildings. The addition of on-site renewables and their accompanying intermittency and non-coincidence requires even greater dynamic performance from the distributed power and energy system. Load following generators, energy storage devices, and predictive energy management are increasingly important to achieve the simultaneous goals of increased efficiency, reduced emissions, and sustainable economics. This paper presents two optimization strategies for the dispatch of a multi-chiller cooling plant with cold-water thermal storage. The optimizations aim to reduce both costs and emissions while considering real operational constraints of a plant. The UC Irvine campus micro-grid operation between January 2009 and December 2013 serves as a case study for how improved utilization of energy storage can buffer demand transients, reduce costs and improve plant efficiency. A predictive control strategy which forecasts campus demands from weather predictions, optimizes the plant dispatch, and applies feedback control to modify the plant dispatch in real-time is compared to best-practices manual operation. The dispatch optimization and predictive control algorithms are shown to reduce annual utility bill costs by 12.0%, net energy costs by 3.61%, and improve energy efficiency by 1.56%.