3896 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 57, NO. 12, DECEMBER 2010 An Online Control Algorithm for Application of a Hybrid ESS to a Wind–Diesel System Chad Abbey, Member, IEEE, Wei Li, Student Member, IEEE, and Géza Joós, Fellow, IEEE Abstract—Energy storage systems (ESSs) can be applied to mitigate some of the negative impacts associated with a variable power generation source such as wind energy. The control of ESS power must be accomplished over numerous time frames to meet system objectives and respect ESS capacity constraints. This paper proposes a two-level ESS control structure for use with a wind–diesel system, which is suitable for online implementation. The control is developed to coordinate power delivered from the two ESS levels in order to minimize diesel fuel consumption and limit up/down rates of the diesel plant. Different control modes are evaluated by simulation, and a subset of the results are validated using a hardware-in-the-loop representation. The con- troller that combines all three functionalities—minimizing dump load, limiting intrahour diesel ramp rates, and maximizing ESS utilization—demonstrates superior performance as measured by defined metrics and is proven to work online. Index Terms—Energy storage, hardware in the loop (HIL), power electronics, power generation, wind energy. NOMENCLATURE C intra (·) Cost of energy associated with simulation of two- level storage system. e mt,i Instantaneous energy state of medium-term stor- age device. e st,i Instantaneous energy state of short-term storage device. E dump (·) Dump load energy function in per unit, expressed on a base of total load energy. E mt Energy rating of medium-term energy storage sys- tem (ESS). E o Initial energy state of ESS. E st Energy rating of short-term ESS. i Index of intrahour time periods running from 1 to I . k Index of hours running from 1 to K. p diesel,i Diesel generator power at time i. p diesel,est Estimated instantaneous diesel power. p diesel,ref ,k Reference power of diesel plant for hour k. p dump,i Dump load power at time i. p dump,est Estimated instantaneous dump load power. Manuscript received April 16, 2009; revised August 17, 2009; accepted December 19, 2009. Date of publication August 26, 2010; date of current version November 10, 2010. C. Abbey is with the Department of Electrical Equipment, Hydro-Québec Research Institute, Varennes, QC J3X 1S1, Canada. W. Li and G. Joós are with the Power Engineering Research Laboratory, De- partment of Electrical and Computer Engineering, McGill University, Montreal, QC H3A 2A7, Canada. Digital Object Identifier 10.1109/TIE.2010.2051392 p ess,ref ,k Reference power of ESS for hour k. p L,i Load power at time i. p mt,i Instantaneous power delivered by medium-term ESS device. p mt, mod Modified instantaneous power reference of medium-term ESS device. p res,i Difference between load and wind power or resid- ual load at time i. p st,i Instantaneous power delivered by short-term ESS device. p st, mod Modified instantaneous power reference of short- term ESS device. p w,i Wind power at time i. P mt Power rating of medium-term ESS. P st Power rating of short-term ESS. P min Diesel minimum power constraint. T mt Time constant for medium-term ESS device in a two-level controller. π e Price of energy supplied by diesel power, in dol- lars per kilowatthour. π ess,e Cost of storage energy capacity, in dollars per kilowatthour per day. π ess,p Cost of storage power capacity, in dollars per kilowatt per day. π w Price of energy supplied by wind power, in dollars per kilowatthour. I. I NTRODUCTION C OMPLEMENTING wind energy with an ESS can lead to improved power generation characteristics. The sources of energy storage may include batteries, [1], supercapacitors, [2], or even the wind turbine inertia, [3], [4]. Coordination with the wind energy resource leads to improved management of the energy delivered, insofar as undesirable frequencies of variation can be selectively attenuated. Numerous papers have considered the scheduling of ESSs on an hourly basis [5]–[7]. The problem is generally formulated as an optimization problem, and detailed modeling of the storage device and wind energy system are not considered. In [8], scheduling of distributed generation (DG) is approached in a similar manner but formulated as a stochastic optimization problem to capture the impact of random variables. The results include the schedule for the ESS or DG over the period in question, operated to either minimize system cost or maximize the revenue of the combined wind–ESS plant. While interesting, hourly scheduling results neglect the im- pact of intrahourly dynamics. These factors (dynamics of the 0278-0046/$26.00 © 2010 IEEE