Runtime optimization of hybrid energy source from an UPS back-upping critical consumer Elena Dănilă, Dorin Dumitru Lucache Faculty of Electrical Engineering, Technical University ”Gheorghe Asachi” of Iasi, Romania
[email protected],
[email protected] Abstract- The vulnerability of electricity consumers must be decreased by integrating specific energy storage systems, which can be used in any situation. As energy distribution and consumption are not constant, the storage units (for backup) must be as high capacity and total availability to streamline the complete system. This paper proposes an objective function of charging an UPS integrated hybrid source (battery and supercapacitor) and a method for optimizing the supply of loads in a critical place of consumption through the UPS with hybrid structure, increasing the autonomy of backup source and its time of re-entry into service.
I.
INTRODUCTION
Optimization can be qualified as the procedure of determining the best solution for mathematical defined problems, which are mostly models of physical reality. It involves study of optimality criteria, determination of solutions with algorithmic methods, the study of the problem’s structure and a computed comparison between real and determined data. Stating the problem always involves finding a balance between building a model complex enough to describe as right the problem and the ease of solving it. Most optimization models have limitations that must be known by those who use them. The first difficulty is that the optimization model is rigid in a certain sense [1] because it implies knowledge of all elements or assumptions of real model. The weakening of these assumptions leads to other types of optimization models as: stochastic optimization, parameter optimization, coefficient optimization or simulation models. A simulation model does not calculate what one should do to achieve some particular purpose but clarify what would happen in a given situation [2]. The purpose of a simulation model can be to forecast the future behavior of the system in certain circumstances, or the design of a policy direction to develop strategies and organizational structures and evaluate their effect on system behavior. In other words, optimization models are prescriptive [3] and the simulation models are descriptive. Energy storage in a power system can be defined as any state function of the system, by which it is possible to store the energy generated, to retain it and use in case of need [4]. According to this definition, energy storage systems (as backup sources, UPS) can be employed in the following states: storing, charging and discharging. In each of them has to be maintained a balance between power output and energy required by the system, so that the storage unit (the source) must have a power capacity of appropriate values.
II.
OBJECTIVE FUNCTIONS OF A STORAGE SYSTEM
Optimization is always relative to one or more implicit or explicit criteria, directed to specific objectives, determining the value system which guides the material efforts, namely investments: 1. storage cost (minimum); 2. the quality (maximum), as a result of the specification on manufacturing tolerances, the chemical composition and nature of the components; 3. performance (maximum), described by operational safety, interchangeability, independence, endurance, availability; 4. adaptability to changes in the demand for energy consumption (maximum); 5. charge cycle duration (minimum) and discharge (maximum), the entry into service (minimum); 6. effectiveness given as the ratio between the yield of use and the factors that contributed to its achievement (maximum). Since a storage system operating optimization is not a problem that can be reduced to a global solution, given the objectives to be met simultaneously, the optimization function is a multiobjective one, associated to dynamic programming. This technique relies from Bellman optimality principle: “if a given state-action sequence is optimal, and we were to remove the first state and action, the remaining sequence is also optimal (with the second state of the original sequence now acting as initial state)” [1]. A general problem of multiobjective optimization can be described as a vector function f that maps a tuple of m parameters (decision variables) to a tuple of n objectives. Mathematically be written: min/max y=f(x)=(f1(x), f2(x),…, fn(x)) (1) where x=(x1, x2,…, xm) ∈ X and y=(y1, y2,…, yn)∈ Y . x is called decision vector, y is called the objective vector, X is the set of decisions and Y is the set objectives. The set of decisions of a multiobjective optimization problem consists of all decision vectors that cannot get to any objective improvement without degrading another objective these vectors are known as Pareto optimal. In other words, all solutions can be optimal when all objectives listed above are met. All decision vectors that are not dominated by any other vector x are called non-dominant (or Pareto optimal) and the set of all alternatives non-dominant are called Pareto Optimal Front. Exploring the space of Pareto optimal solutions in a
reasonable time to obtain a final solution is made by evolutionary algorithms [5]. The most used methods to solve multiobjective problems in modern electrical engineering are: A. Algorithm OEGADO (Objective Exchange Genetic Algorithm for Design Optimization) reduces the number of evaluations needed to find a Pareto optimal region by applying reduced models [6]. OEGADO run several concurrent instances of genetic algorithms, optimizing each instance of an object and forming a scale model of its objective. At fixed intervals every algorithm changes its reduced model with others [7]. B. If certain applications of energy storage systems objectives can be compared, ranked, the method of parameter optimization (or weights method) is applied. It transforms the case into a single-objective problem by combining all individual objective functions into a single function with a single solution [8]. C. Another approach to the multiobjective problem, also within the hypothesis of transforming it in a model with a single objective function, is given by the global criterion method or utopia point method [9]. D. The goal attainment method (developed by Gembicki), applied in this paper for optimizing the operating time of a backup source, is a method that solves the issue with a vectorvalued index function. It uses vector optimization as a tool for analyzing static control problems with performance and parameter sensitivity indices [10]. The method involves the following steps: the decision maker attaches an aspiration level to each objective (aspiration level of an objective is a specific value (realistic) to achieve the aim, different from the restrictions of the problem). Optimization criterion is reduced to minimize the sum of absolute values of differences between the actual values and aspiration levels: n
min(∑ f k ( x) − Fk ,
(2)
1
where Fk is the aspiration level associated with the objective function [11]. In addition to the levels Fk, the decision maker has to define a weight vector w (which shows the relative preference for each of the function’s objectives), associated with over achievement or failure of Fk. The sum of the vector’s elements (weights) is 1. ⎧min z , (3) ⎪⎪ ⎨ f k ( x) − z ⋅ wk ≤ Fk , for f k ( x) → min ⎪ ⎪⎩ f k ( x) + z ⋅ wk ≥ Fk , for f k( x ) → max The level of achieving the objectives is appreciated by a real variable z (if z is positive greater than 1 there is a failure, and if it is negative there is an over achievement) [12]. Since the storage system is reliable if provides energy with specific parameters in admissible intervals (voltage, power) and within precise time intervals (charging / discharging, during entry-into-service), any optimization method must be completed, for relevance, with boundary conditions.
III.
AUTONOMY IMPROVEMENT OF UPS SOURCES BY SUPERCAPACITORS INTEGRATION
The most common backup sources for critical consumers are online double conversion Uninterruptible Power Sources. The scale of use is varied, from modular UPS (in applications requiring easy expansion and fast maintenance), conventional UPS (with microprocessors for precise, constant control of all measurements and power factor correction) to line-interactive UPS (for individual workstations, switchboards, home automation applications and for small service sector companies). These units must cover the need of load in case of power line disturbances. The most frequent power-quality events (according to IEC 62040-3) are: outages (>10ms), voltage fluctuations (
