Modelling and simulation of a gri...
P177, Page 1 8th International Conference on System Simulation in Buildings, Liege, December 13-15, 2010 Modelling and simulation of a grid connected photovoltaic heat pump system with thermal energy storage using Modelica R. De Coninck1,2*, R. Baetens3, B. Verbruggen4, J. Driesen4, D. Saelens3, L. Helsen1 (1) Division of applied mechanics and energy conversion section, Department of mechanical engineering (2) 3E, BE-1000 Brussels, Belgium (3) Division of building physics, Department of civil engineering (4) Electrical energy computer architectures, Department of Electrical engineering (1)(3)(4) K.U.Leuven, BE-3000 Leuven, Belgium ABSTRACT When the penetration of renewable electricity production in the electricity infrastructure increases, an increased part of the production follows a stochastic behaviour. In order to reduce grid peak loads and to maintain the required balance between production and consumption at all times, two solutions can be envisaged: electricity storage and demand side management (DSM). One typical DSM solution consists of using thermal energy storage (TES) to decouple electric loads from thermal demands. In order to study the dynamic interaction between thermal (incl. building) and electric systems, their integration in one single simulation environment is required. This study develops a model in the object oriented language Modelica and uses the model to assess the impact of additional TES capacity. The model describes an energy concept consisting of a dwelling with a grid connected photovoltaic system, a compression heat pump, a hot water storage tank and a control strategy. The multi-disciplinary model, developed in this study, is a first step towards the simulation of complex systems in which thermal and electric components are combined in order to study for example the effects of DSM on grid stability. The results of a simulation study, using the newly developed model, are presented. The benefits of adding thermal storage capacity with regard to the overall seasonal performance factor (SPF) and the impact of the system on the electrical grid are analysed for a standard control strategy and two variants: a control strategy focusing on operation during daytime and a control strategy focusing on limiting net power exchange peaks. The daytime strategy is able to increase the overall SPF for different storage tank sizes if the storage tank is sufficiently insulated. Both alternative control strategies are able to substantially reduce the number of net power exchange peaks, even with relatively small storage tanks. Keywords: Photovoltaic, heat pump, thermal energy storage (TES), grid load, simulation, Modelica 1. INTRODUCTION On May 18th 2010 the European parliament adopted a recast of the Directive on Energy Performance of Buildings (2002/91/EC - the Directive is expected to be published in the official journal in June 2010, the version adopted by the European Parliament on the 23rd of
P177, Page 2 8th International Conference on System Simulation in Buildings, Liege, December 13-15, 2010 April 2009 can be found in (The European Parliament, 2009)). Article 9 of this Directive obliges EU member states to build only ���near zero energy buildings��� (near ZEB) from 2020 onwards. Although the definition of a near ZEB in the EU Directive is not elaborated, many different definitions for ZEB can be found in the literature. Torcellini et al. (2006) discuss the impact of four different definitions and conclude that the choice of definition in the design phase influences the energy concept of the building. However, for each of the definitions, it is possible to reach the ZEB target by a combination of energy efficiency, heat pumps and sufficient photovoltaic (PV) systems. In such an all-electric building, the PV system has to cover the electricity consumption on a yearly basis in order to be a site, source or emission ZEB. Only a cost ZEB ��� for which the net yearly energy services bill has to be zero - might require more PV production than the yearly consumption, depending on the electricity tarification. From this analysis it can be expected that buildings with a heat pump and a photovoltaic system will become standard practice in new constructions in the short to medium term. Already today we see a strong growth on the domestic heat pump and PV markets (European Heat Pump Association, 2010), (EurObserv'ER 2009). The major flaw of each of the definitions investigated by Torcellini et al. (2006) is the yearly basis for the analysis. If a large share of the buildings would be ZEB with PV and heat pumps installed, the impact on the electricity grid could be substantial: all these buildings would inject electricity on the grid when the local production exceeds consumption and take electricity from the grid in the opposite cases. These time periods characterized by either peak injection or consumption would occur simultaneously for the majority of these buildings as the weather conditions (both solar radiation and temperature) dictate to a large extent both the electricity production (via PV) and consumption (via the heat pumps). This simultaneity can cause grid stability problems, as described in different contributions (Pepermans, Driesen, Haeseldonckx, Belmans, & D, 2005), (Vu Van, Woyte, Soens, Driesen, & Belmans, 2003), (Houseman, 2009). The specific case of grid coupled PV with a heat pump heating system has been simulated by Baetens et al. (2010). In this paper, solutions to reduce the grid impact of a combined PV and heat pump concept for a single family dwelling are investigated, while keeping track of the heat pump system performance. The paper focuses on the influence of the size of the storage tank and the control strategy for the heat pump with the aim to increase the SPF of the whole system (and thus lowering total energy use) and reduce both the size and amount of net power exchange peaks. 2. MODEL DEVELOPMENT A detailed model has been developed in Modelica (The Modelica Association, 1997). Modelica is an open source, object oriented and equation based modeling language. Modelica offers the advantage that the differential and algebraic equations (DAE) that describe the physical behaviour of the components are solved in one DAE system instead of solving all components sequentially. The object oriented approach also enables an easier integration of previous modelling work. Through the use of Optimica, Modelica offers extended functionality with regard to (dynamic) system optimisation (��kesson, ��rz��n, G��fvert, Bergdahl, & Tummescheit, 2009). The model is schematically presented in Figure 1. It consists of a 2-zone building, an air-to- water heat pump, a stratified storage tank, a heat distribution system with two radiator circuits
P177, Page 3 8th International Conference on System Simulation in Buildings, Liege, December 13-15, 2010 and a grid connected PV system. The different components of the model are described in sections 2.1 to 2.4. Figure 1 ��� General model scheme 2.1 Building In this paper, a high-order lumped capacitance model (Clarke 2001, Underwood & Yik 2004) for predicting the unsteady building response is developed within Modelica. A first-order lumped capacitance approach is used for each of the different construction layers of the building, whereas a second-order model is used for all surface layers in order to cope with restrictions for low Biot-numbers (Incropera et al. 2007) and to allow a more accurate prediction of indoor surface temperatures. Similarly as in the TRNSYS type56, inter-surface longwave radiation is modeled by means of a zone star temperature (Davies 1993), reducing the complexity of the model compared to view factors and allowing a straight-forward implementation of internal convective and radiative gains. The absorption, transmission and reflection of solar radiation through the array of air and glass layers in the windows is modeled with the embedded technique described by Edwards (1977, 1982). The transmitted diffuse short-wave solar radiation is distributed over all room surfaces weighted to their surface and absorption coefficient, whereas the direct short-wave solar gains are modeled to fall on the floor. Liesen & Pedersen (1997) show that only small differences would arise when making different assumptions on the distribution of the transmitted solar energy. Within the developed building model, a stochastic generated occupancy profile from Richardson et al. (2008) and domestic load profile are included. The modeled domestic electricity consumption profile takes into account standby power (de Almeida et al. 2008) and domestic cooling appliances (Firth et al. 2008, Liu et al. 2004), lighting (Stokes et al. 2004, Richardson et al. 2009), fan operation, cooking (Glorieux & Vandeweyer 2002, Wood & Newborough 2003) and the use of media like television and computer (Glorieux &