Environmental Engineering Reference
In-Depth Information
uncertainty about whether wind speeds will be high enough to cause turbine pro-
tection to activate, and a subsequent reduction in wind power production. Implementation options
It is clearly sensible for system operators to follow the fuel saver approach initially,
since, at low penetration levels, wind generation can be accommodated econom-
ically and technically without too much forethought. However, with a growth in
wind farm development, and increased confidence in wind forecasting technology,
system operators will undoubtedly drift towards the latter wind forecasting
approach, reflecting ever increasing confidence in the predicted wind profiles. It is
worth noting, however, that demand forecasting is also a probabilistic process and
assumes that, although the behaviour of an individual consumer cannot be foreseen,
the likely behaviour of different categories of customer can be predicted with some
confidence. It is therefore simplistic to present the switch from fuel saver mode to
wind forecast mode as involving a transition from a deterministic problem to a
probabilistic one. At all times the system operator must still ensure that the demand
is met and that sufficient fast reserve response and other ancillary services are
available. Clearly, the wind forecasting approach is more challenging to implement
and will require the system operator to modify and expand his thinking, but the
environmental and other benefits are obviously significant.
In order to support understanding of the wind profile at a particular operational
time, TSOs require that regular data be supplied increasingly for individual wind
farms using SCADA systems. For example, a requirement for real and reactive
power production is common (Energinet, Eirgrid, E.ON, SvK). Meteorological data
is required by some TSOs, including wind speed (Energinet, Eirgrid), in addition to
wind direction, ambient temperature and ambient pressure (Eirgrid). Control status
information (SvK, Energinet, Eirgrid) may include available capacity, curtailment
setpoint (on/off), regulation capability (on/off) (SvK), percentage shut down due to
high wind speed and islanding detection. Prior to its merger with the NETA
(new electricity trading arrangements) market, the requirements in Scotland were
perhaps the most stringent, with data required on most of the above, and also fre-
quency control status (on/off) and power system stabiliser (PSS) functionality.
The data, and in particular any meteorological information, can be used to
inform the output of wind forecasting tools. A wind power prediction tool (WPPT)
is employed in Denmark, with updated forecasts provided every 6 hours for the
next 48 hours in 1 hour steps. Wind speed, wind direction, air temperature and
power output from reference wind farms in each sub-area are provided as inputs to
the system, to enhance the accuracy of the forecasts (Holttinen, 2005c). An
advanced wind power prediction tool (AWPT) also uses a statistical model, in
conjunction with data from 25 wind farms and measurements from 100 single
turbines. In the E.ON area, AWPT achieves typically a 6 per cent average (RMS)
error for 6 hours ahead and a 10 per cent error between 24 and 48 hours ahead
(Ensslin et al. , 2003). For the future, 'Zephyr' has been proposed as a merger of
'Prediktor' (physical) and WPPT, both outlined in Chapter 6, using data from all
Danish wind farms rather than representative sites (Nielsen et al. , 2002). Although
Search WWH ::

Custom Search