|
| |
Further Discussion
Electric load varies significantly across the day, week, month, and season,
and the effects of other significant influences (eg weather) vary by time of day
and date (i.e. day of the week, holiday versus workday, etc). The TESLA model is
tailored to the particular patterns experienced in each franchise area,
including any tendency of those patterns to shift.
The human behaviour that underlies variation in load varies in regular patterns
that can largely be predicted based on knowledge of the clock and calendar.
Variables that are time- and date-based are therefore excellent predictors of
load. The TESLA model has internal facilities to determine the day of the week,
holidays or other days of special significance, annual seasonal patterns, and
the proximity of a given day to some day of special significance. For example,
Wednesday load obviously differs from Sunday load, but it also differs from the
other days of the week, as well. Behaviour and load are different on and around
holidays, and the difference depends on the specific holiday and the day of the
week on which it occurs.
The daily cycle explains much of the variation in load, and that cycle differs
based on the day of the week and proximity to a holiday. In identifying
holidays, the model is locale-specific. For example, Boxing Day (26 December) is
observed in the UK and Canada, but not in the US. Cinco de Mayo is a holiday in
Mexico, but not in the US; however, it would qualify as a "special day" in many
parts of the Southwestern US.
There are also strong interactions between clock and calendar variables and the
weather. For example, the load on a rainy Sunday will be much different than the
load on a rainy Tuesday, and the effects of the rain itself on load will be
different on a Sunday, compared to what it is on any other day of the week. The
relationship between load and temperature also varies dramatically based on
clock and calendar effects. The model explicitly captures these relationships
and interactions, through a richly specified set of variables that capture
daily, seasonal, and special day effects.
|
