Wind power and electricity markets - Wind Power Integration: Connection and System Operational Aspects - page 254

scales - minutes, hours and days - requires that there is a plan/schedule in place
to commit/de-commit other plant (typically thermal) to maintain the balance. The
greater the amount of wind power installed, the larger the error in the expected
value. Also, the longer the time horizon of the wind forecast, the greater the error.
Shorter time frames for forecasting will reduce the error to very small amounts, but
this will require flexible plant to operate the system. Load forecasting also intro-
duces a degree of unpredictability but is not as sensitive to time horizon, and for
large wind penetrations the wind forecast error will predominate over the longer
time horizons. Clearly the matching thermal generators need to have characteristics
that make them suitable to follow this variability and/or unpredictability. Under the
causer pays principle, any additional costs incurred in providing these balancing
services need to be borne by the wind power generators. Costs of increased wind
generation include: capital cost of flexibility in matching thermal plant, extra start-
ups and shutdowns and ramping of conventional plant (Denny et al. , 2006).
Operation of a power system with significant amounts of wind power is a
unique challenge that will require technical and market innovation. The design and
development of these new electricity markets are heavily influenced by the
operational characteristics of the electricity system. Wind power has characteristics
that make it significantly different from the more traditional generation technolo-
gies, in particular its variability. Variability of the wind power is the expected
change in output over a time horizon; the error in this expectation is a measure
of the unpredictability. In order to extract maximum benefit from installed wind
power it may be necessary to modify the existing operational methods (Chapter 5),
and these modifications need to be reflected in the corresponding market structure.
For example, unit commitment has traditionally being formulated as a deterministic
optimisation problem. This was based on the assumption that the forecasts for load
and unit availability were generally accurate. With wind power it may be appro-
priate to formulate the unit commitment problem as a stochastic optimisation
problem as the wind forecast error can be large. Failure to modify and/or adapt
operational methods may hinder the development of wind power. Through the
market the costs of these modifications and additional services need to be allocated
effectively.
System operations and electricity market arrangements need to be closely
aligned, i.e. the economic and engineering principles need to be coordinated in
order to optimise the electricity industry. There are some who believe that certain
operational and market practices are not favourable to wind power and have put
forward arguments in favour of altering market rules and/or operational practices.
For example it can be argued that a long gate closure time, e.g. 12 hours, is detri-
mental to wind power, as the forecast error can be very large over this time frame
(30% - Chapter 6). This exposes wind operators to major buy and sell volumes in
the balancing mechanism. The price in the balancing mechanism will typically be
high when wind is short and low when wind is long. Shorter gate closure times will
reduce the volumes that are exposed to the balancing mechanism. This argument
was successfully used to shorten the gate closure time in the predecessor to the
BETTA market from 4 to 1 hour (OFGEM, 2002). However, the fundamental issue