Maximizing storage flexibility in an aggregated heat pump portfolio

Abstract

To balance the electrical grid due to a large increase in the renewable energy production mainly from wind turbines will be a problem in the near future in Denmark. Smart grid solutions with new storage capacities are essential. In this work single family houses with heat pumps are investigated for storage capability. It is of great importance to move energy consumption in time to balance the grid. In this paper a portfolio of houses are modelled and controlled using an aggregated model and a model free scheduling algorithm. Flexibility and ability to trade on the intra-day regulating market is increased by use of a novel algorithm. I. INTRODUCTION In Denmark at present wind power meets 30 % of the electricity demand, however this covers variation from a minimum of 2-3 % to peaks more than 100 % of the instantaneous power demand. The Danish government wants the wind energy percentage to be larger than 50 % in 2020, (1). Wind power production gives balancing problems in the grid and ways to overcome this are investigated in a number of projects. In this work the increasing number of heat pump heated single family houses are used in a smart grid solution. These installations have large energy storage capacities in the concrete floors giving a possibility of moving energy consumption to times where a high production of wind energy takes place with only minor discomfort and thereby they are very flexible power consumers. Single family heat pumps have been investigated in several works (2), (3), (4). The idea has been to make central control based on predicted market prices, weather forecasts and models of the individual houses. The control strategy is based on MPC acting on single houses. This concept is sensitive to a correct house model. In (5) it is shown that it is difficult to make a dynamic model of inhabited houses making it possible to predict power consumption based on in and outdoor temperatures, solar radiation and thermal house parameters as the inhabitants causes large disturbances. Therefore it is chosen not to use individual models but to use an aggregated model for a pool of houses. It is assumed that the parameter variations and disturbances on the individual houses may be evened out adding the input and outputs of the individual models. The major part of the electricity in Denmark is traded on the NORD POOL Elspot market, (6), which is a day-ahead market. The intra-day unbalances must be treated by buying *This work is a part of READY! , a project supported by PSO funds administrated by Energinet.dk via the ForskEl project program 2012-1- 10757