Representation of wind power generation in economic models for long-term energy planning

Abstract

The representation of wind power plants electricity generation in economic models for energy planning is problematic, since generation in wind power plants is variable and it is not possible to predict accurately enough wind fluctuations for more than a few days. Often, wind power plants generation patterns from a single historical year are repeated throughout the modelled time period. Typically, this method is used when analysing power system operation in hourly time intervals for all days in a year for each year. However, hourly time resolution is not feasible in large-scale models, which tend to require considerable amounts of computing power. Thus, some kind of time aggregation is needed. On the other hand, currently used methods for models with less than hourly temporal resolution are becoming inadequate because of increasing share of fluctuating electricity production from wind in total power generation.In this article, a methodology for evaluation of wind power plants stochastic electricity generation in power system development models is described. The methodology is based on evaluation of how much time single or multiple wind power plants generate a certain output range during a season or some time period within a year, modelling of output distribution for a typical day of the selected time period, and preparation of electricity output curves. These electricity output curves when modelling wind power plants in models with aggregated time allow to assess fluctuations in generation, observable regularities and enhance the objectivity of balancing power demand assessment, also ensure that electricity generated during a typical day corresponds to electricity generated during the selected time period. This methodology will help to determine the rational perspective power generation mix more accurately and make a better assessment of cost-effectiveness of wind power plants in economic models for energy planning, without significantly increasing the size of already large-scale models.

A Review on Time Series Aggregation Methods for Energy System Models

“…Among the early model frameworks, one group focuses on long-term system planning and has usually only one time step per year such as LEAP [43], EFOM [44], and BESOM [45], which are not subject to aggregation techniques and thus neglected in the following. The temporal dimension of the other major group of early bottom-up ESMs such as TIMES [46][47][48][49] and its predecessors MARKAL [50], MESSAGE [12,51], IKARUS [9], and PERSEUS [52] are based on time slice formulations (in the case of PERSEUS called "time slots"), which are explained in more detail in Section 3.2.1. Although the long-term planning models with only one time step per year were consecutively combined with models with a higher temporal resolution, as is the case for TIMES [46][47][48][49] as a combination of MARKAL [50] and EFOM [44], the time slice approach, which is based on the modeler's experience only, remained unchanged for decades.…”

“…To the best of our knowledge, the idea of time slices (TSs) was first introduced by the MESSAGE model [9,51] and the expression was reused for other models, such as THEA [104], LEAP [105], OSeMOSYS [106], Syn-E-Sys [107], and TIMES [48,49]. The basic idea is comparable to that of averaging, but not based on aggregating consecutive periods.…”

“…Instead, the method is sometimes simply called "representative day" [66,67,[108][109][110][111][112][113], "TD" [54,[114][115][116][117][118][119][120][121], "typical daily profiles" [16,17], "typical segment" [122] "time slot" [52], or "time band" [123]. Accordingly, the term "TS" is used by the majority of authors [36,39,51,[104][105][106][107][124][125][126][127][128]. The most frequent distinction is made between the four seasons [16,17,36,39,[104][105][106][107]115,121,124,[126][127][128] or between summer, winter.…”

“…The most frequent distinction is made between the four seasons [16,17,36,39,[104][105][106][107]115,121,124,[126][127][128] or between summer, winter. and mid-season [40,51,54,66,67,91,[108][109][110]112,117,118,120,123,129,130], but other distinctions such as monthly, bi-monthly, or bi-weekly among others [40,51,54,111,113,114,116,119,122,125] can also be found. Within this macro distinction, a subordinate distinction between weekdays and weekend days [16,17,51,106,107,111,113,116,…”

“…and mid-season [40,51,54,66,67,91,[108][109][110]112,117,118,120,123,129,130], but other distinctions such as monthly, bi-monthly, or bi-weekly among others [40,51,54,111,113,114,116,119,122,125] can also be found. Within this macro distinction, a subordinate distinction between weekdays and weekend days [16,17,51,106,107,111,113,116,[121][122][123], weekdays, Saturdays, and Sundays [115,124,126], Wednesdays, Saturdays, and Sundays [104,105], or others, such as seasonal, median, and peak [40] can be found. In contrast to the normal averaging, each TS does not follow the previous one, but is repeated in a certain order a certain number of times (e.g., five spring workdays are followed 13 times by two weekend spring days before the summer periods follow).…”