Distributed Photovoltaics in the
Swedish Energy System. Model Development and Simulations.
Uppsala University, (2009)Abstract
Application of photovoltaics (PV) is increasing worldwide, mainly due to extensive subsidy
schemes for introducing CO2 -free power generation. A majority of newly installed systems
are distributed small-scale systems located in distribution grids, often at residential customers.
Recent developments suggest that such distributed PV generation (PV-DG) could gain more
interest in Sweden in the near future. With prospects of decreasing module prices, an extensive
integration could be possible.
This licentiate thesis presents the first part of a PhD project with the aim to determine the
potential for domestic PV-DG in Sweden. Two aspects are treated in detail in the thesis: (1) the
ability of PV to match a local domestic power demand and (2) impacts of extensive integration
of PV-DG on power flow in low-voltage (LV) distribution grids. To make realistic studies for
high-latitude conditions, there is a need for representative demand and PV generation data. As
there is a lack of detailed domestic load data in Sweden, a major part of the work has been
devoted to development of a stochastic load model.
Interdisciplinary studies of household activities were performed to get insight into how do-
mestic electricity use is embedded in the structure of everyday life. It was found that time-use
(TU) data, normally used in the social sciences, can be used to model domestic power demand.
Both a conversion model for estimating power demand from empirical TU data and a stochastic
Markov-chain model for generating synthetic activity patterns and power demand were devel-
oped and extensively validated against measurements. Importantly, a realistic model of domestic
lighting demand from occupancy patterns and irradiation data was developed, that preserves the
negative correlation between irradiation and lighting demand. The models provide a basis for
load matching studies and power-flow simulations, but can be used for other purposes as well.
Case studies of individual households showed that the appearance of daily load profiles, and
thus the degree of matching to PV generation, are highly variable. Studies of matching of PV
generation to aggregate domestic demand showed that load matching at moderate overproduc-
tion levels can be improved by PV panel orientation, demand side management (DSM) and
storage. At high overproduction levels, however, the only impacting option is storage.
Probabilistic power-flow simulations with the developed models yield a versatile picture
of how impacts are distributed among customers and over time, as compared to often-used
static simulations. Contrary to the trend towards higher time resolution in international research,
hourly resolution was found to be sufficient for determining probability distributions for LV grid
voltages. Power-flow simulations of three Swedish LV grids showed that a penetration level of 1
kWp PV systems at every customer was most beneficial in terms of on-site coverage of demand,
counteracted voltage drops and decreased network losses. However, much higher penetration
levels, up to the highest studied level of 5 kWp per household, can be handled without voltage
rise above prescribed limits.