C I R E D 22 nd International Conference on Electricity Distribution Stockholm, 10-13 June 2013 Paper 1440 CIRED2013 Session X Paper No 1440 Determination of load schedules and load shifting potentials of a high number of electrical consumers using mass simulation Johannes BRAUNAGEL, Petrit VUTHI, Wolfgang RENZ, Holger WIECHMANN Hans SCHÄFERS, Hojat ZARIF EnBW Vertrieb GmbH - Germany Center for Demand Side Integration, h.wiechmann@enbw.com University of Applied Sciences Hamburg - Germany Johannes.Braunagel@haw-hamburg.de ABSTRACT The massive growth in the usage of renewable, but fluctuating energy sources in connection with the energy turnaround calls for a similar growth of the integration of flexible loads. This seems to be a vital aspect to maintain the balance between production and consumption in the transportation grid as well as in the distribution grid at low voltage level. In this regard the question of how to achieve a flexible and cost efficient mass integration of low power demand facilities like domestic heat pumps is of special interest. The Center for Demand Side Integration at the University of Applied Sciences Hamburg and Energie Baden- Württemberg (EnBW, the 3rd largest utility in Germany) developed and tested control algorithms in a simulation project in order to integrate a large number of distributed consumers (several 10.000) into an integrated system. The behaviour of the system was analysed with regard to load prediction, scheduling and shifting potentials. INTRODUCTION The electrification of heat supply in residential buildings and therefore the connection of electricity and heat demand cannot only increase the efficiency regarding primary energy consumption, but also holds the possibility to integrate these consumers as “smart” applications into the electricity supply system. Electrical heating systems like heat pumps (HP) or night storage heaters (NSH) can have a big flexibility to postpone their electrical consumption. Due to installed storage systems and/or the storage capacity of the in-house heat distribution system the electricity demand can partly be decoupled from the heat demand. Heating systems in the residential sector, especially heat pumps, only have small installed loads but can be aggregated to switchable loads of a significant dimension. In order to enable an operator to take control of a swarm consisting of a large number of small consumers, special algorithms need to be developed which enable the aggregation of the flexibility in power consumption and the monitoring of the system. Also thermal marginal conditions have to be met to ensure that no loss of comfort arises due to too high or too low room temperature in the buildings or a lack of hot tap water. SYSTEM DESIGN AND ELEMENTS The objective in the development of the system design was to create a structure that enables efficient communication and a central control and monitoring of a system for Demand Side Integration, which consist of a large number of system elements (tens of thousands). An additional requirement was the day ahead scheduling of the consumers participating. In order to optimize the procurement of energy at a wholesale market, a day-ahead load schedule needs to be generated for all of the consumers. It had to be taken into account that a demand-dependant planning makes the knowledge of the demand behaviour and the state of all the systems elements mandatory. To collect and evaluate this data centrally would lead to an immense data volume. A further aspect is the scheduling problem itself. Often used methods like LP (Linear Programming) or MILP (Mixed Integer Linear Programming) can be used for the scheduling of single or comparably smaller groups of applications. The speed in finding the optimal solution and the computational effort depends on the constraints, the number of applications and the objective function. Especially the dispatching of a high number of applications with similar parameters leads to a flat solution tree which slows down the finding of the optimal solution dramatically [1]. A central approach for this scheduling problem would not be feasible regarding the computational overhead or at least limit the scalability of the system. On the basis of these considerations an agent based concept was used where decentred control and scheduling are realized. This concept bears additional advantages regarding the reliability and extendibility of the system. During a failure of communication an autonomous agent can fall back on a load schedule that was generated earlier or an emergency schedule. New participants can easily be integrated without changing the central data stock. Components of the system and system architecture The top-level entity of the system is formed by the dispatcher. Analogous to power plant dispatching the dispatcher optimises the operation of the whole system, gives guidance for the generation of the day-ahead load schedule and is able to change the behaviour of the swarm within the day. Based on external information such as shortages in the distribution grid or current prices at the