Title: Global energy management system for cooperative networked residential green buildings
Abstract: IET Renewable Power GenerationVolume 10, Issue 8 p. 1237-1244 Research ArticleFree Access Global energy management system for cooperative networked residential green buildings Hanane Dagdougui, Corresponding Author Hanane Dagdougui [email protected] Department of Electrical Engineering, École de Technologie Supérieure (ÉTS), Montréal, CanadaSearch for more papers by this authorAhmed Ouammi, Ahmed Ouammi National Center for Scientific and Technical Research (CNRST), Rabat, MoroccoSearch for more papers by this authorLouis Dessaint, Louis Dessaint Department of Electrical Engineering, École de Technologie Supérieure (ÉTS), Montréal, CanadaSearch for more papers by this authorRoberto Sacile, Roberto Sacile Department of Informatics, Bioengineering, Robotics and Systems Engineering, Faculty of Engineering, Genoa, ItalySearch for more papers by this author Hanane Dagdougui, Corresponding Author Hanane Dagdougui [email protected] Department of Electrical Engineering, École de Technologie Supérieure (ÉTS), Montréal, CanadaSearch for more papers by this authorAhmed Ouammi, Ahmed Ouammi National Center for Scientific and Technical Research (CNRST), Rabat, MoroccoSearch for more papers by this authorLouis Dessaint, Louis Dessaint Department of Electrical Engineering, École de Technologie Supérieure (ÉTS), Montréal, CanadaSearch for more papers by this authorRoberto Sacile, Roberto Sacile Department of Informatics, Bioengineering, Robotics and Systems Engineering, Faculty of Engineering, Genoa, ItalySearch for more papers by this author First published: 26 July 2016 https://doi.org/10.1049/iet-rpg.2015.0282Citations: 14AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract This study addresses an optimisation problem faced by a network of green buildings (NGBs) connected to the main electrical grid. The problem is formulated as a cooperative internal power control among interacting residential buildings. The cooperation is reached through a communication infrastructure in the NGB, where the global central controller of the network is the responsible for the coordination of the local residential buildings' energy management systems by properly allowing the optimal management of the internal and external power flows in each building. The main advantage of the cooperation among residential buildings is to better match the load profile of each building internally (at the network level). In order to achieve this goal, components such as energy storage system, distributed generations and loads are included. The uncertainties characteristics of wind speed, solar irradiation, and loads are also considered for the control and operation of the whole system. A small network of five residential buildings has been simulated using the proposed model. Numerical results demonstrate the effectiveness of the proposed network. Nomenclature Variables and parameters β(ψ ) beta probability density function for the solar radiation Γ gamma function ψ solar radiation (kW/m2) probabilistic solar radiation (kW/m2) ξ, ϕ parameters of the β(ψ ) probabilistic output power of the PV modules at the GB i at instant t (kW) Apv solar cell array area (m2) ηpv PV module reference efficiency Pf packing factor ηpc power conditioning efficiency f(ν ) Weibull probability function k, c parameters of the f(ν ) ν wind speed (m/s) probabilistic wind speed at the GB i at instant t (m/s) probabilistic output power of the wind turbine at the GB i at instant t (kW) Pr rated power (kW) vc cut-in wind speed (m/s) vr rated wind speed (m/s) vf cut-out wind speed (m/s) Si,t state variable expressing variation of ESS from the reference value in GB i at instant t (kWh) θij,t control variable related to power exchange between GBs i and j at instant t (kW) θi,net,t control variable related to power exchange between GBs i and the main grid at instant t (kW) desired power exchange between GBs i and j at instant t (kW) desired power exchange between GBs i and the main grid at instant t (kW) γi weighting factor related to the cost of an exceeding/lacking quantity of energy stored in the ESS i to the reference value weighting factor relate to the cost of the power exchange with the other buildings weighting factor relate to the cost of power exchange between each building i and the main grid predicted energy consumption vector at the GB i (kWh) Xi energy consumption vector at the GB i (kWh) accepted bounds of fluctuation of the energy consumption at the GB i (kWh) φi total energy consumption at GB i (kWh) yi,t state variable related energy stored in each GB i (kWh) total probabilistic energy generated at the GB i at instant t (kWh) predicted energy demand of GB i (kWh) αi efficiency of the ESS at GB i limit related to the energy consumption at the GB i (kWh) Si,max upper bound related to ESS in the GB i (kWh) Si,min lower bound related to ESS in the GB i (kWh) 1 Introduction Recently, the amount of research related to smart grids has been growing rapidly. Smart grid is considered as the future of power grid, which is able to manage the production, transmission and distribution of electricity in an intelligent manner. However, the transition to smart grids needs to be done gradually, and may appear in medium to long-term as a network of microgrids (MGs) [1]. Under the smart grid and the MGs paradigms, the role of buildings in the electric power grid can move from a passive to an active participant in order to manage grid and serve as a key player in maintaining the grid reliability. In this case, the building may be considered as a generating-resource or/and consuming-resource depending on the status of power generation and consumption in the building [2]. The main advantage that a building integrated MG offers is to enable customers to have a bidirectional communication platform that allows them to send, receive, save, and even control their energy needs and excesses. The interest on residential building integrated MG has generated a high interest on the use of renewable energy sources (RES) and distributed generators. Specifically, renewable energy-based building integrated MG can be taken into account as a better way for exploiting renewable energy, enhancing energy efficiency, decreasing electricity bills and reducing the use of fossil fuels. The capability of smart residential buildings to construct a cooperative network is the further development and application of the concept of MG and smart buildings. In this perspective, connecting smart residential buildings can provide a nice real-world opportunity for testing the efficient building operation and supporting each residential green building (GB) to its neighbouring GBs for a sustainable energy use. However, new control challenges and complexities that regard power control and reliability arise due to the coordination between GBs and with the main grid. In addition, the uncertainties introduced by the RES and load consumption make it more difficult to realise optimal energy management. A number of papers on smart residential buildings in the literature have focused on cost reduction, energy management strategies and load management through demand response programs. Kanchev et al. [3] presented a determinist energy management system for a MG, including an advanced photovoltaic (PV) generator with embedded storage units and a gas microturbine. Molderink et al. [4] developed a three-step control methodology to optimise the electricity import/export by reshaping the energy profiles of the houses. Dagdougui et al. [5] provided a dynamic model able to integrate different RES and one storage device to feed a building for its thermal and electrical energy needs in a sustainable way. The authors in [6] introduced a dynamic decision model for the real time control of hybrid RES production system. A computationally feasible and automated optimisation-based residential load control scheme in a retail electricity market is presented in [7]. Previous published papers on the subject have taken valuable steps towards realising effective energy management in buildings. However, the management and control were performed from a unitary building viewpoint and the capability to connect many residential buildings to be part of a network and exchange power with neighbouring buildings has not been enough investigated. Ouammi et al. [8] studied the optimal energy management in a network of MGs based on the use of the original linear quadratic Gaussian problem. The main drawback of the approach is related to unbounded character of the state and control variables resulting from the optimisation problem. Dagdougui et al. [9] solved a similar problem using an original modelling approach based on the mathematical formalisation of the Pontryagin minimum principle. However, these approaches are either computationally intensive and/or not suitable for real-time applications, or can produce suboptimal solutions. Arefifar et al. [10] proposed a planning model to divide a distribution system into networked MGs for its optimal self-healing. Recently, the authors in [11] proposed an optimal planning of MGs interconnection which considers a clustering method for representing the deployment of variable renewable energy. In [12], Wang et al. proposed a decentralised power dispatch model for the coordinated operation of multiple MGs and a distribution system. The problem is based on a decentralised bi-level stochastic optimisation algorithm with the main grid in the upper level and MGs in the lower level. The energy storage systems (ESSs) are not included in the considered network of MGs. Kumar Nunna and Doolla [13] proposed a multiagent system-based energy management system to control the operation of networked MGs and allow customers to participate in demand response. However, it can be seen that the concept of using local building renewable generation and storage system to support other buildings of network of GBs (NGBs) are not considered in the above. In addition, these studies do not consider the study of cooperation among GBs in the paradigm of GBs network, especially with the consideration of developing a global control strategy for the exchange of information and forecast of power production and consumption on the whole set of GBs in the network. The main focus of this paper is to propose a cooperative centralised control algorithm for the control of a NGB, where power lines can be established among different buildings, and with the main electrical grid. The GBs considered in this paper are not only consuming electricity, but are also capable of generating and storing it using their local renewable power generation and electric storage system. We see these buildings as green as they can generate power and consume it from the use of the local available RES. Since the supply–demand balance can always be met for buildings connected with a utility system, the objectives of the GBs are to minimise the buildings' electricity buying by optimally scheduling the charge/discharge of energy in its local energy storage and by optimally exchanging power with its neighbouring GBs. In this paper, the control is implemented in a centralised way, where a global centralised controller (GCC) is coordinating the NGBs. The GCC is provided with relevant information from each building energy management system (BEMS) regarding local forecasted loads, wind and solar forecasting, and operating limits of various systems in order to find appropriate scheduling and coordination of local resources. By integrating intelligent control algorithms with information technologies, the GCC can control the energy charge/discharge in each ESS and power distribution in each building, while taking into account also their participation in an external distribution network. It is assumed that adequate communication infrastructure is available which guarantee the efficient bidirectional information exchange between the BEMS and GCC. Compared to the previous literature review, the contribution of the paper is twofolds: We present a global energy management system for cooperative networked residential GBs. Each building is a RES-based GB modelled as a system including loads, ESS and producing renewable energy from wind and solar resources. This paper contributes to demonstrate the advantages of cooperation among a set of GBs by exploiting uncertainties of renewable sources and demands. We present a detailed model for each ESS in the GB, which takes into account constraints related to the storage system. In addition, the energy consumed in each building is constrained by a certain level of acceptance that is the bounds which the decision maker (DM) is satisfied with. The problem is formulated as a cooperative internal power control among interacting residential buildings. The main challenge is the development of a centralised algorithm for the smart management and control of the power exchanges in a network of residential buildings able to cope with dynamics and uncertainties of renewable production and loads. A case study is adopted to simulate the real practices and to test the proposed approach. Numerical studies demonstrate the usefulness and efficacy of the proposed model. Moreover, we show that the proposed framework can account for different practical constraints. The rest of the paper is organised as follows: Section 2 introduces the concept of cooperation between the GBs and Section 3 describes the model of the GB. Section 4 illustrates the mathematical model of the proposed NGB. In Section 5, the results of a case study are presented. Finally, the paper concludes with Section 6. 2 Design and concept of cooperation of GBs 2.1 GB description Smart buildings have attracted lots of attention in recent years. Recent papers [14, 15] have focused on the control and optimisation methods applied to the smart houses. Fig. 1 shows the GB electrical system architecture. As for the MG, the GB can be connected or disconnected from the main grid and/or other neighbouring buildings. In general many GBs' configurations can be found, alternating current (ac), which is the most common over the world, direct current (dc) and mixed ac/dc. Commonly, ac distribution presents a certain number of benefits: ac voltage can be incremented and decremented easily through the use of electrical transformers and circuit protection is more mature for ac systems. While dc distribution presents a certain number of benefits, such as lower impedance, nonreactive power flow on line, and rectification stages are avoided and only smaller dc/dc conversion stages are needed [16]. In addition, dc distribution systems offer several advantages such as the easiness integration of some RES and batteries which can be easily connected to dc distribution systems. Fig. 1Open in figure viewerPowerPoint Electrical system architecture of an AC GB The GB consists of an integrated hybrid system in which the distributed energy resources operate in the form of a grid that can be connected to or disconnected from the neighbouring grids (other GBs and/or the main grid). We consider a GB as a small internal grid equipped with PV panels, wind turbine, electrical ESS and load. The GB is also equipped with converters, that are assumed to be coordinately controlled to supply an uninterrupted power under variable conditions for solar irradiation, wind speed and loads fluctuations, thus, when the building is connected to other grids or in isolated modes. 2.2 Centralised control of NGBs The approach considered in this paper is related to a global centralised control of a NGB, where an interconnected number of buildings are cooperating. At the NGB level, a GCC is available and is responsible for the coordination of the BEMSs by properly allowing the optimal operation of each building, while managing the relation between the building and its external distribution network (main grid and other GBs) as well as with other GBs to which it is connected. This topology allows each building to sell/purchase electricity to/from other GBs, to charge/discharge energy in the local ESS, and to sell/purchase electricity to/from the main grid. The BEMS available in each GB implements a set of functions such as load, wind and solar power predictions, customers' preferences, and overall security assessment [17]. Through proper communication structure, each BEMS transfers a set of relevant building information based on all information collected from sensors and smart meters and its processed forecasting systems (local load, wind speed, solar irradiance) to the GCC. In addition, common external information are gathered by advanced metering infrastructure from the distribution network, and which can be, for example related to energy price forecasting in a day-ahead market. Fig. 2 shows the conceptual architecture of the NGB, and the interactions between the GCC and the BEMSs. Fig. 2Open in figure viewerPowerPoint Conceptual architecture of the NGB In view of the above discussions, the GCC may have the privilege to manage large scale of information gathered from different GBs of the network, and find the priority of assigning a giving generated power by assigning either a higher priority local load utility, higher priority neighbouring load utility, higher priority grid utility or higher priority local storage utility. In the proposed model, we are focusing on non-flexible loads, so in this case, high priority is given to ensure the buildings loads, before planning any benefits from the power surplus. 2.3 Modelling the NGBs Let's consider a network of buildings. The topology of the network is represented by a directed graph , where denotes the cardinality of the set of GBs, and denotes the cardinality of the set of power lines. We note that power exchange can take place in both direction and the row direction is purely conventional. The centralised approach proposed is fully cooperative, since all BEMSs send information on their state, forecasted production and demand to the GCC, which can compute and forward the optimal decisions to each GB, including the optimal flow exchanged on each existing line and the charge/discharge of ESS in each GB. 3 GB components 3.1 PV modules generator The electrical energy generated by PV modules is mostly affected by some characteristics of the site such as solar irradiance and ambient temperature, and other characteristics of the module itself. The solar irradiance can be modelled by the Beta probability density function using historical data to describe the random phenomenon of the irradiance data; it is given by [18, 19] (1) with The parameters can be computed using the mean () and the standard deviation (σ) of the solar irradiance as follows [18, 19]: (2) The probabilistic output power of the PV modules at the ith GB is calculated as follows [15]: (3) 3.2 Wind turbine generator The Weibull probability function is used to represent the frequency of the wind speed, and it also represents the most frequent starting point of stochastic analysis, simulation and forecasting of wind speed. Its general formulation is represented as follows [20]: (4) In the literature, numerous models of the output power of wind turbines are available. In the current work, a simplified linear model is used; it assumes a linear dependence of the wind turbine power output on the current wind speed at the hub height. The probabilistic power output of the wind turbine is as follows [21]: (5) with (6) 3.3 Energy storage system A smart GB is expected to be equipped with an ESS, such as batteries, hydrogen storage system or plug-in hybrid electric vehicles and so on. In this paper, the ESS is modelled as an energy reservoir with a certain degree of performance and upper and lower storage bounds. 4 Mathematical modelling 4.1 Multi-objective optimisation The objective to be minimised is characterised by the sum of three terms that are properly weighted in one function: (i) the quadratic deviation of the stored energy from a reference value; (ii) the quadratic deviation of power exchange of each GB with other buildings from a desired reference value; and (iii) the quadratic deviation of power exchange of each GB with the main grid from a desired reference value. The decisions variables Si,t are introduced, defining the amount of energy in the storage device at GB i ∈ I. In addition, the decisions variables θij,t and θi,net,t are introduced, defining the exchange of power within each GB (7) More specifically, these expected desired reference values can be obtained following a statistical analysis of previous historical data. In this paper, it is assumed that the reference values are a priori known by the DM. The weighting factors do not have monetary values, but they refer to a priority in the choice of interconnection with respect to another. It is worthwhile to underline that the weighting parameters in the cost function play a key role in the resulting solution. So, when the computational effort allows the computation of several solutions, a sensitivity analysis on these parameters should be performed in order to give a clear idea of the robustness of the solution to the DM, showing trade-off among objectives and Pareto optimality. This aspect is shown in the result section. In addition, some specific techniques, such as analytical hierarchy process, may be adopted to allow the DM to evaluate the importance of each objective with respect to the others, and as a consequence the weights in the cost function, according to several criteria. From a mathematical viewpoint, the problem proposed in this paper is a non-linear quadratic problem. Despite its formulation according to mathematical programming, it is a model allowing further solving techniques as linear quadratic tracking (through relaxing some of its constraints), MPC based tracking or using the Pontryagin's principle. 4.2 Constraints and state equations of the loads The energy consumption scheduling for each GB i (i ε I) is defined by a vector Xi = [xi,1, xi,2, …, xi,T] and is the predicted energy consumption vector, where is the predicted energy demand of the ith GB in time interval (t, t + 1]. The total energy consumption during a time horizon is given as follows: (8) The energy consumed in a GB is constrained by a certain level of acceptance. It means the bounds which the DM is satisfied with (9) where is a reference value, it characterises the accepted bounds of fluctuation of the energy consumption. For the satisfactory operation of the GB, load management of each building must be incorporated into the optimisation problem of the whole system of GBs. This ensures that appropriate load shedding takes place when required in order to prevent the system against a total failure and to provide a secure power supply to those loads with high utility. We note that it is assumed that each building has a certain need to be assured in a desired time interval [tm, tn] as follows: (10) The energy consumed at each time interval in a building i has a certain limit imposed by the utility as follows: (11) 4.3 Constraints and state equations of the ESS The energy stored in each GB is assumed to be described by the following state equation: (12) δij is a binary variable that defines the state of the line established with other buildings, δi,net is a binary variable expressing whether a communication link is established with the building i and the main grid or not. For example, a value '0' for all the buildings means that the network is operating in an autonomous way (13) (14) The state variable in each building is yi,t (kWh) which represents the quantity of energy stored in the local ESS at time t. This value can be also expressed as a variation from the reference value of the ESS in the ith GB, (kWh). The state equation of the ESS can be rewritten introducing the following change of the state variable: (15) (16) (17) where μi,t is a known value in time interval, it represents the probabilistic energy balance which means the difference between the probabilistic generation and the consumption (18) (19) The energy stored is constrained by an upper and lower bounds (20) 5 Application and numerical results 5.1 Simulation setup This section provides numerical analysis to demonstrate the performance of the presented optimisation algorithm. It is worthwhile to mention that the optimisation problem presented in the previous sections is solved by LINGO software. In the adopted network, each building is connected to two adjacent buildings and to the main grid via power lines (PLs). The proposed NGBs is shown in Fig. 3. Each GB is assumed to be equipped with two distributed renewable generators (roof-top solar arrays and wind turbine). The capacity of the ESS in each GB is bounded between a minimum of 1 kWh and a maximum of 15 kWh. In this simulation, the efficiencies of the storage devices of the five GBs are considered to be equal respectively to 90, 60, 80, 70 and 85%. Fig. 3Open in figure viewerPowerPoint Network architecture of GBs In addition, for the five GBs, it is supposed that the costs of power exchange with the buildings and the main grid are a priori known. The simulation is run for 24 time slots; each slot having different local power balance in each GB. It is worth to mention that when solving the proposed, prediction tools are needed to obtain the price of power exchanges, the wind and solar power generations and the loads. However, as the focus of this paper is not prediction methods, the predictions are assumed to be perfect. In this case, the forecasts of renewable power generations and loads are made by some appropriate methods. Their algebraic sum results in a forecasted power balance (μi,t) characteristic of each GB. The forecast μi,t is reported in Table 1, it is represented by independent distributed random vectors. Table 1. Forecasted values of μi,t (kW) for the GBs GB1 GB2 GB3 GB4 GB5 1 1.93 −2.33 5.95 4.74 1.19 2 1.57 7.27 3.84 3.99 2.20 3 2.09 −1.29 −1.81 2.39 −0.94 4 −1.21 3.60 0.67 5.95 −1.06 5 −3.86 6.39 −0.12 1.28 4.07 6 −0.11 −0.66 −4.02 2.61 −3.28 7 5.72 −5.56 3.21 1.15 2.87 8 −4.23 −1.78 4.47 1.48 3.40 9 1.98 0.84 −5.16 1.24 0.99 10 −1.30 −0.21 0.78 −2.39 −2.82 11 0.79 2.32 1.79 −0.37 4.66 12 −1.02 0.15 0.74 −3.47 −5.46 13 −5.11 −0.99 2.29 3.12 −4.97 14 3.12 3.83 1.92 2.13 2.66 15 −0.28 4.70 −2.65 −2.48 3.01 16 0.97 −1.63 −3.60 4.63 8.34 17 1.17 2.78 5.40 2.00 4.50 18 −9.01 −0.66 2.77 0.37 2.78 19 3.98 6.50 5.59 −1.01 1.98 20 −4.16 2.44 3.11 2.37 1.51 21 0.60 1.27 −0.08 0.21 2.58 22 2.94 −0.22 −2.62 4.42 −0.85 23 5.25 −2.90 0.56 −3.58 6.34 24 −3.78 0.23 0.62 0.57 −1.95 It is also assumed that the GBs are dimensioned to be autonomous both as a network and on their own. So , and are a null vectors, while the incoming/outgoing power exchange expected to be less than 5 kW (θi,j,t < 5 and θi,net,t < 5) is allowed on each power line. 5.2 Solving process of the optimisation problem According to quadratic programming techniques, the global optimum is found using quadratic recognition pre-processor of LINGO software that is able to automatically determine whether an arbitrary non-linear programming is a convex and quadratic. This recognition allows then passing to the faster quadratic solver, that is available as part of the barrier solver. When this latter is combined with the global option, LINGO software will automatically recognise second-order cone models, in addition to convex quadratic models and then will dramatically generate great speed advantages for large-scale problem. For the centralised scheme of NGBs, central processing unit (CPU) time and convergence speed rate are of paramount importance for real-time implementation. In this paper, we have implemented the model using LINGO 14.0 optimisation software in an AMD Phenom™ II X4 B95 Processor 3.00 GHz with 4 GB RAM, it takes few seconds for Lingo software to schedule the power exchanges under the centralised control scheme. 5.3 Results and discussion Fig. 4 compares the amount of power exchanged internally between the GBs considering fixed internal and external exchange costs of . For instance, the solid line gives the power exchanged among GBs 2 and 3, where θ23 can take positive (power is directed from GB 2 to GB 3) and negative values (power is directed from GB 3 to GB 2) at different times. It can be seen that maximum power exchange in the GBs network is reached for θ45. It can be seen that at t = 7, a maximum value of 0.86 kW is sent from GB 3 to GB 2. Globally, the figure shows that the power is mainly exchanged internally at the network level. This outcome reasonably verifies the results with the proposed cooperative centralised control approach. Here, we should mention that the power exchange control depends on the power
Publication Year: 2016
Publication Date: 2016-07-26
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
Cited By Count: 18
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot