Environmental Engineering Reference
In-Depth Information
It will also be shown that the correlation of the produced wind power diminishes
and the predictability of wind power increases as the wind generation capacity grows.
Then it becomes beneficial to optimise a system by defining and applying cost func-
tions rather than optimising forecasts on the mean absolute error (MAE) or the root
mean square error. This is because the marginal costs of up and down regulation are
asymmetric and dependent on the competition level of the reserve market. The advan-
tages of optimising wind power forecasts using cost functions rather than minimum
absolute error increase with extended interconnectivity, because this serves as an im-
portant buffer not only from a security point of view, but also for energy pricing.
1.
Introduction
Wind power is considered one of the most important renewable energy sources in the
near future. In the past years the number of wind farms world wide increased strongly,
with a total installed capacity of more than 94 GW by the end of 2007 [1], with about 50%
distributed in Germany, Spain and the USA. This high concentration suggests that from a
global perspective there is space for ample additional wind power. The issue is therefore not
whether it is feasible, but rather what does it cost? The main focus in wind power integration
in the past has been on producing the most accurate forecast with minimal average error, but
experience has shown that this is not necessarily optimal from a cost perspective. Balancing
costs will increase with increasing volumes of wind power. The larger the forecasted error
(in MW), the higher the balancing costs.
Another aspect is the concentration level of wind power, which has a side effect from
a forecasting perspective. This is the correlated generation and forecast error and is mainly
relevant for large amounts of offshore wind power, as it is planned e.g. in the North Sea
to in order to reduce transmission costs. Such scenarios make it unfeasible to run an elec-
tricity network with a forecast optimisation target of minimal mean absolute error, because
it implies short periods during which GW of fossil fuel based power plants will have to
be started with short notice. Germany is such an example, where it is the large errors that
dominate the balancing costs of wind power. Skewness of the balancing costs of negative
and positive reserve gives advantage to conservative forecasts in such areas and requires
curtailment or other scheduled plants to stop generation with short notice. Such scenar-
ios are also regularly experienced in Spain. An EU 6th-Framework project is assisting the
Transmission System Operators (TSO's) in developing tools to handle such cases with a
so-called cluster management [2].
From these experiences, it appears that cost functions will not only benefit the market
prices and balancing costs, but also act as a means to increase system security. For those
countries, where the installed capacity has gone beyond the 10 GW level, it has become an
important consideration in the daily operation.
Whether intermittency poses technical limits on renewables in the future is certainly
also of concern for other forms of renewable energy sources [3], since OECD (Organisa-
tion for Economic Cooperation and Development) Europe and IEA's (International Energy
Agency) World Energy Outlook [4] project up to 23% market share of non-hydro renewable
energy by 2030. Natural variations of resource availability do not necessarily correspond
with the (also varying) need of the consumers. Balancing supply and demand is therefore a
Search WWH ::
Custom Search