J. Widen1*, E. Wackelgard1 and P. Lund2

1 Department of Engineering Sciences, Uppsala University, P. O. Box 534, 751 21 Uppsala, Sweden
2 Advanced Energy Systems, Helsinki University of Technology, P. O. Box 2200, FI-02015 HUT, Finland

Corresponding Author, ioakim. widen@angstrom. uu. se

Abstract

Extensive integration of distributed photovoltaic (PV) generation into residential areas is likely to be a challenge to the control of power quality in low-voltage distribution grids. Overproduction of power caused by mismatching load and demand profiles puts restrictions on the amount of distributed generation allowed locally in the grid at high penetration levels. This paper investigates the load matching capability of PV at Stockholm, Sweden, and how it can be improved by different measures; PV panel orientation, demand side management (DSM) and electricity storage. Detailed models for PV generation and household electricity load are used. Energy storage has the greatest potential of obtaining a better match between load and production in terms of solar fraction at system setups where the production is several times higher than the mean load, although both DSM and PV array orientation options have comparable impacts. At lower levels of overproduction, orientation and DSM options seem to be slightly more effective, mainly because of energy losses in the storage medium.

Keywords: Photovoltaics, Distributed generation, Load matching

1. Introduction

Utilization of solar energy is gaining increasing interest as a means of reaching sustainability in the built environment. In Sweden, and at high latitudes in general, direct generation of electricity with solar photovoltaics (PV) has traditionally been ascribed a low potential for contribution to the total domestic electricity demand. As recent studies have pointed out, however, grid-connected distributed PV, e. g. roof-top systems, could gain attention as an energy-saving measure in domestic buildings [1].

In Sweden, therefore, it seems likely that a more extensive integration of PV will start at the lowest voltage level of the distribution grid. This makes the load matching capability of the electricity production a critical issue. If production and load do not match, overproduction of power could lead to problems with power quality at high penetration levels of distributed generation, of which the main issue is voltage rise [2]. There are some theoretical studies on this effect for high latitudes (see for example [3], [4]) but almost no measurements or observations, because the critical amount of generation is yet to be integrated.

In this paper three options for improved load matching are studied. The first option is orientation of the photovoltaic arrays. For maximization of the array output at any instant of time, the incidence angle of the incoming radiation should be zero. For maximum annual production at high latitudes,
the optimal azimuth angle is due south, and the optimal tilt somewhat higher than the latitude. If matching with the evening or morning load is desirable, however, orientation eastward or westward could be an option, favouring morning and evening sun respectively.

The second option is altering the demand profile through demand side management (DSM). This refers to the active rescheduling of energy-demanding activities or appliances in the households. From a utility’s point of view, a smoothened load curve is desirable, which could be achieved by encouraging households to reschedule energy use to low-demand periods. For better load matching with photovoltaics, loads should instead be shifted toward mid-day.

These two options are compared to shifting of production to high-demand periods through energy storage. A description of the models used for generating electricity load and production data follows in Section 2. Section 3 describes the different load matching options in more detail and results of simulations for improved matching with household electricity load in detached houses are presented in Section 4.

2. Models

For the end-use-specific household electricity demand, a previously developed load model was used, validated and adjusted against preliminary measurements from the Swedish Energy Agency (SEA). For generation of photovoltaic electricity generation data, a detailed calculation procedure was applied to hourly series of direct and diffuse irradiation on the horizontal plane.