Environmental Engineering Reference
In-Depth Information
incident (see Figure 3.10, discussed later) therefore
K
is not related to rotating inertia effects.
The value of K for a specifi c power system depends on the governing characteristics of
the generators and the combined frequency-power sensitivity of the loads. Its value varies
during the day as the share of load may shift between frequency sensitive and insensitive
consumers.
The value of
K
can be estimated by tripping out a large generator and measuring the result-
ing frequency drop after the transients have subsided. Preferably, values of
K
can be derived
at random times from the records of system response after the loss of a generator due to a
fault. To ensure stable operation after the most severe contingency that the utility considers
as
credible
, i.e. probable,
K
should be such that the system frequency does not drop below a
value that leads to serious underperformance of power station auxiliary drives. Of course, the
utilities cannot protect the system from contingencies that have very low probability, e.g. the
simultaneous loss of two large power stations. If such events were to occur, measures are in
place to maintain the integrity of the system by means of unavoidable power interruption to
some consumers, but more of this will be given later.
3.4 Dynamic Frequency Control of Large Systems
3.4.1 Demand Matching
Figure 3.2 showed the daily variation of the aggregate load for a typical utility. The pattern
of low load at night and the peaks in the morning and afternoon is a feature of such demand
curves. Superimposed on this daily cycle are faster random variations.
Power systems operation is conventionally broken down into different timescales ranging
from seconds to days. During second to minute load variations partly loaded plants respond
through governor action. Generating plants responding in this timescale are said to be provid-
ing
continuous service
or
frequency response
.
The next time scale is
load following
, which involves the connection/disconnection of plant
to balance the anticipated load increases/decreases. This timescale covers approximately 10
minutes to several hours during which decisions are taken in response to the trend in demand
on the basis of plant operation economics. For example, in anticipation of the early morning
increase in demand, the system operator either directly or through bidding mechanisms
(Chapter 7) is responsible for ensuring that suffi cient capacity is available to meet these large
but relatively slow and predictable changes in demand.
Fossil fuelled generators, and especially plant fuelled by coal, can be readily made to follow
demand. Such generators are brought on line to supply any escalating demand according to
cost effectiveness and fl exibility. This results in the layering of generation shown in Figure
3.7. Nuclear power being infl exible occupies the bottom of the pile. CCGT generation and
effi cient coal occupy the next layer. These two layers supply what is known as the
base load
with other coal fi red generation and imports from adjacent networks required to do the load
following. Pumped hydro, when availble, is used to extract energy when demand is low and
to deliver it during peak demand.
For load following to be effective it is essential to have accurate predictions of the demand
curve. Programming generation at this timescale is known as
unit commitment
or
generation
scheduling
.
